torch.ones(0) does not work.

## 🐞Describing the bug

## Stack Trace

```
Converting PyTorch Frontend ==> MIL Ops: 86%|████████▌ | 6/7 [00:00<00:00, 3241.77 ops/s]
Running MIL Common passes: 0%| | 0/37 [00:00 NeuralNetwork Ops: 100%|██████████| 2/2 [00:00<00:00, 32768.00 ops/s]
/Users/ryosukefukatani/work/HMERModel/venv/lib/python3.9/site-packages/coremltools/models/model.py:149: RuntimeWarning: You will not be able to run predict() on this Core ML model. Underlying exception message was: Error compiling model: "Error reading protobuf spec. validator error: LoadConstantNDLayer 'var_8' has empty constants.".
_warnings.warn(
Traceback (most recent call last):
File "/Users/ryosukefukatani/work/HMERModel/atnBTTR/d17.py", line 19, in
out_dict = model_ct.predict({'x': x.detach().numpy().astype(numpy.float32)})
File "/Users/ryosukefukatani/work/HMERModel/venv/lib/python3.9/site-packages/coremltools/models/model.py", line 545, in predict
raise self._framework_error
File "/Users/ryosukefukatani/work/HMERModel/venv/lib/python3.9/site-packages/coremltools/models/model.py", line 147, in _get_proxy_and_spec
return (_MLModelProxy(filename, [compute_units.name](http://compute_units.name/)), specification, None)
RuntimeError: Error compiling model: "Error reading protobuf spec. validator error: LoadConstantNDLayer 'var_8' has empty constants.".
```

## To Reproduce

```python
import torch
import coremltools as ct
import numpy

class Net(torch.nn.Module):
def forward(self, x):
return torch.ones(0)

torch_model = Net()
x = torch.randn(3)
traced_model = torch.jit.trace(torch_model, x)

model_ct = ct.convert(traced_model,
inputs=[ct.TensorType(shape=x.shape, name='x')])

out_dict = model_ct.predict({'x': x.detach().numpy().astype(numpy.float32)})
```
## System environment (please complete the following information):
- coremltools version 6.0:
- MacOS: 12.4

Start with the provided Python reproduction, especially the torch.ones(0) return value and the ct.convert call. Trace how this empty tensor is represented during conversion, then run the example through model_ct.predict. Done means the conversion and prediction no longer fail with an empty constants validation error.

Written by the indexing model from the issue text.