PR CI builds failure due to incompatibility of recent protobuf with pytorch

Recent CI PR builds are failing due to incompatibility of recent `protobuf` (Python `protobuf` `4.*`) with proto files in pytorch.
Previously, `tensorboard` was the one to constrain `protobuf` to `<4`, but since https://github.com/tensorflow/tensorboard/pull/6147 was merged (released with `2.12.0`) - it supports `protobuf` `4.*` and version specifier is relaxed to include `4.*`.

[Change announcement in protobuf](https://protobuf.dev/news/2022-05-06/#python-updates).

## Environment

detectron2 PR CI

## __Full logs__ or other relevant observations

Occurs in #4938.

error log from CI

```
============================= test session starts ==============================
platform linux -- Python 3.8.6, pytest-7.3.1, pluggy-1.0.0 -- /opt/circleci/.pyenv/versions/3.8.6/bin/python3.8
cachedir: .pytest_cache
rootdir: /home/circleci/detectron2
plugins: hydra-core-1.3.2
collecting ...
collecting 15 items
collecting 56 items / 4 errors
collected 297 items / 6 errors

==================================== ERRORS ====================================
_________________ ERROR collecting tests/test_export_caffe2.py _________________
tests/test_export_caffe2.py:18: in
from detectron2.export import Caffe2Model, Caffe2Tracer
detectron2/export/__init__.py:9: in
from caffe2.proto import caffe2_pb2 as _tmp
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/__init__.py:11: in
from caffe2.proto import caffe2_pb2, metanet_pb2, torch_pb2
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/caffe2_pb2.py:33: in
_descriptor.EnumValueDescriptor(
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/google/protobuf/descriptor.py:796: in __new__
_message.Message._CheckCalledFromGeneratedFile()
E TypeError: Descriptors cannot not be created directly.
E If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
E If you cannot immediately regenerate your protos, some other possible workarounds are:
E 1. Downgrade the protobuf package to 3.20.x or lower.
E 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).
E
E More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
__________________ ERROR collecting tests/test_export_onnx.py __________________
tests/test_export_onnx.py:11: in
from detectron2.export import STABLE_ONNX_OPSET_VERSION
detectron2/export/__init__.py:9: in
from caffe2.proto import caffe2_pb2 as _tmp
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/__init__.py:11: in
from caffe2.proto import caffe2_pb2, metanet_pb2, torch_pb2
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/caffe2_pb2.py:33: in
_descriptor.EnumValueDescriptor(
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/google/protobuf/descriptor.py:796: in __new__
_message.Message._CheckCalledFromGeneratedFile()
E TypeError: Descriptors cannot not be created directly.
E If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
E If you cannot immediately regenerate your protos, some other possible workarounds are:
E 1. Downgrade the protobuf package to 3.20.x or lower.
E 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).
E
E More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
______________ ERROR collecting tests/test_export_torchscript.py _______________
tests/test_export_torchscript.py:17: in
from detectron2.export import dump_torchscript_IR, scripting_with_instances
detectron2/export/__init__.py:9: in
from caffe2.proto import caffe2_pb2 as _tmp
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/__init__.py:11: in
from caffe2.proto import caffe2_pb2, metanet_pb2, torch_pb2
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/caffe2_pb2.py:33: in
_descriptor.EnumValueDescriptor(
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/google/protobuf/descriptor.py:796: in __new__
_message.Message._CheckCalledFromGeneratedFile()
E TypeError: Descriptors cannot not be created directly.
E If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
E If you cannot immediately regenerate your protos, some other possible workarounds are:
E 1. Downgrade the protobuf package to 3.20.x or lower.
E 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).
E
E More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
________________ ERROR collecting tests/test_model_analysis.py _________________
tests/test_model_analysis.py:8: in
from detectron2.utils.analysis import find_unused_parameters, flop_count_operators, parameter_count
detectron2/utils/analysis.py:10: in
from detectron2.export import TracingAdapter
detectron2/export/__init__.py:9: in
from caffe2.proto import caffe2_pb2 as _tmp
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/__init__.py:11: in
from caffe2.proto import caffe2_pb2, metanet_pb2, torch_pb2
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/caffe2_pb2.py:33: in
_descriptor.EnumValueDescriptor(
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/google/protobuf/descriptor.py:796: in __new__
_message.Message._CheckCalledFromGeneratedFile()
E TypeError: Descriptors cannot not be created directly.
E If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
E If you cannot immediately regenerate your protos, some other possible workarounds are:
E 1. Downgrade the protobuf package to 3.20.x or lower.
E 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).
E
E More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
__________________ ERROR collecting tests/export/test_c10.py ___________________
tests/export/test_c10.py:8: in
from detectron2.export.c10 import Caffe2RPN
detectron2/export/__init__.py:9: in
from caffe2.proto import caffe2_pb2 as _tmp
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/__init__.py:11: in
from caffe2.proto import caffe2_pb2, metanet_pb2, torch_pb2
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/caffe2_pb2.py:33: in
_descriptor.EnumValueDescriptor(
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/google/protobuf/descriptor.py:796: in __new__
_message.Message._CheckCalledFromGeneratedFile()
E TypeError: Descriptors cannot not be created directly.
E If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
E If you cannot immediately regenerate your protos, some other possible workarounds are:
E 1. Downgrade the protobuf package to 3.20.x or lower.
E 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).
E
E More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
_________________ ERROR collecting tests/modeling/test_rpn.py __________________
tests/modeling/test_rpn.py:7: in
from detectron2.export import scripting_with_instances
detectron2/export/__init__.py:9: in
from caffe2.proto import caffe2_pb2 as _tmp
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/__init__.py:11: in
from caffe2.proto import caffe2_pb2, metanet_pb2, torch_pb2
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/caffe2/proto/caffe2_pb2.py:33: in
_descriptor.EnumValueDescriptor(
/opt/circleci/.pyenv/versions/3.8.6/lib/python3.8/site-packages/google/protobuf/descriptor.py:796: in __new__
_message.Message._CheckCalledFromGeneratedFile()
E TypeError: Descriptors cannot not be created directly.
E If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
E If you cannot immediately regenerate your protos, some other possible workarounds are:
E 1. Downgrade the protobuf package to 3.20.x or lower.
E 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).
E
E More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
=============================== warnings summary ===============================
detectron2/data/transforms/transform.py:46
/home/circleci/detectron2/detectron2/data/transforms/transform.py:46: DeprecationWarning: LINEAR is deprecated and will be removed in Pillow 10 (2023-07-01). Use BILINEAR or Resampling.BILINEAR instead.
def __init__(self, src_rect, output_size, interp=Image.LINEAR, fill=0):

tests/config/test_instantiate_config.py:18
/home/circleci/detectron2/tests/config/test_instantiate_config.py:18: PytestCollectionWarning: cannot collect test class 'TestClass' because it has a __init__ constructor (from: tests/config/test_instantiate_config.py)
class TestClass:

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
=========================== short test summary info ============================
ERROR tests/test_export_caffe2.py - TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
1. Downgrade the protobuf package to 3.20.x or lower.
2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
ERROR tests/test_export_onnx.py - TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
1. Downgrade the protobuf package to 3.20.x or lower.
2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
ERROR tests/test_export_torchscript.py - TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
1. Downgrade the protobuf package to 3.20.x or lower.
2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
ERROR tests/test_model_analysis.py - TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
1. Downgrade the protobuf package to 3.20.x or lower.
2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
ERROR tests/export/test_c10.py - TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
1. Downgrade the protobuf package to 3.20.x or lower.
2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
ERROR tests/modeling/test_rpn.py - TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
1. Downgrade the protobuf package to 3.20.x or lower.
2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

More information: https://developers.google.com/protocol-buffers/docs/news/2022-05-06#python-updates
!!!!!!!!!!!!!!!!!!! Interrupted: 6 errors during collection !!!!!!!!!!!!!!!!!!!!
======================== 2 warnings, 6 errors in 2.79s =========================

Exited with code exit status 2

```