Improve the warning message regarding local function not supported by pickle

🐛 Describe the bug

I used a local function in a pipeline which results in the warning:

UserWarning: Local function is not supported by pickle, please use regular python function or functools.partial instead.

So I wrapped the local function in functools.partial and the warning persisted, so I investigated closer. The problem appears to be that _check_unpickable_fn is undoing the partial before checking if fn is local, i.e.

This code unwraps partial, converting back to the underlying local function

https://github.com/pytorch/pytorch/blob/301644d3cb8ce7ecb477b5bcb95a7a8a1304fa71/torch/utils/data/datapipes/utils/common.py#L132

And this code then produces the warning

https://github.com/pytorch/pytorch/blob/301644d3cb8ce7ecb477b5bcb95a7a8a1304fa71/torch/utils/data/datapipes/utils/common.py#L136

My calling code is

def run(args)

    def batch_transform(x):
        return {"inputs": text_transform(x["text"]), "target": label_transform(x["label"])}

    train_datapipe,, test_datapipe= DATASETS[args.dataset](root=args.data_dir, split=('train', 'test'))
    
    # shuffle and batch
    train_datapipe = train_datapipe.shuffle().batch(args.batch_size).rows2columnar(["text", "label"])
    train_datapipe = train_datapipe.map(partial(batch_transform))

This produces the warning regardless of whether batch_transform is partial or not (a fix is to import dill or lift batch_transform out of run)

Versions

PyTorch version: 1.13.0+cu117
Is debug build: False
CUDA used to build PyTorch: 11.7
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.1 LTS (x86_64)
GCC version: (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0
Clang version: Could not collect
CMake version: version 3.22.1
Libc version: glibc-2.35

Python version: 3.10.6 (main, Nov 14 2022, 16:10:14) [GCC 11.3.0] (64-bit runtime)
Python platform: Linux-5.15.0-58-generic-x86_64-with-glibc2.35
Is CUDA available: True
CUDA runtime version: 11.7.99
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration: GPU 0: NVIDIA GeForce RTX 2080 Ti
Nvidia driver version: 525.60.13
cuDNN version: /usr/lib/x86_64-linux-gnu/libcudnn.so.7.6.5
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

Versions of relevant libraries:
[pip3] numpy==1.24.1
[pip3] pytorch-ignite==0.4.10
[pip3] torch==1.13.0+cu117
[pip3] torchdata==0.5.0
[pip3] torchtext==0.14.0
[conda] Could not collect

Start in torch/utils/data/datapipes/utils/common.py at _check_unpickable_fn and reproduce the warning with a local function wrapped in functools.partial. Review how partials are unwrapped before the local-function check; done means the warning behavior or wording accurately reflects the callable being passed.