Can not convert identity image model

Can not convert identity image model due to a name sanitization bug.

```python
import torch
from torch import nn
import coremltools as ct

class Model(nn.Module):
def forward(self, x):
return x

m = Model()
m.eval()

x = torch.randn([1, 3, 1024, 1024])
traced_model = torch.jit.trace(m, x)

y = traced_model(x) # This works

mlmodel = ct.convert(traced_model,
convert_to="mlprogram",
inputs=[ct.ImageType(name="imageInput",
shape=x.shape,
color_layout=ct.colorlayout.RGB,)],
outputs=[ct.ImageType(name="imageOutput",
color_layout=ct.colorlayout.RGB,)],
)
```

Produces:
```
File ~/miniconda3/envs/prod/lib/python3.8/site-packages/coremltools/converters/mil/mil/passes/name_sanitization_utils.py:102, in _sanitize_block_input_vars(block, sanitizer_vars, main_input_types, sanitize_main_input_only)
100 for input_name, var in block.inputs.items():
101 msg = "Main block's input name, '{}', is different from its corresponding var's name, '{}'."
--> 102 assert input_name == var.name, msg.format(input_name, var.name)
103 new_name = sanitizer_vars.sanitize_name(var.name)
104 new_input_dict[new_name] = var

AssertionError: Main block's input name, 'colorImage', is different from its corresponding var's name, 'colorOutput'.
```

Reproduce the conversion with the PyTorch example and inspect coremltools/converters/mil/mil/passes/name_sanitization_utils.py, especially _sanitize_block_input_vars. Trace why the main input key is colorImage while its variable is colorOutput; done means the provided ct.convert call completes without this assertion.

Written by the indexing model from the issue text.