Enable dynamo onnx export of BasicUNet

[asfiyab-nvidia]

Enable dynamo onnx export of BasicUNet models

TorchDynamo based ONNX export does not support torch.jit.isinstance(). To enable the export via dynamo, I've updated the calls to torch.jit.isinstance to isinstance.

Note: I've also verified that the TorchScript based ONNX export still works with this change. These changes enable the export of BasicUNet, BasicUNetPlusPlus, FlexibleUNet, TorchvisionFC models.

Types of changes

• Non-breaking change (fix or new feature that would not break existing functionality).
• Integration tests passed locally by running ./runtests.sh -f -u --net --coverage. - Errors caused due to dataloader
• Quick tests passed locally by running ./runtests.sh --quick --unittests --disttests. - Errors caused due to dataloader

[ericspod]

Judging from the errors you do need to specify Dict with the contained types even though you're explicitly checking for these and Python's isinstance won't use them.