added return_grad for all types of rnnt loss

[durson]

Grad tensors can be useful for the full (ordinary and pruned) rnnt losses.

[pkufool]

Can you also fix this issue (#25 (comment)), thanks!

[durson]

I fixed #25 (comment)
However I preserved S > 0 requirement since otherwise it is not compatible with torchaudio and unittest was not passing.

Summary of the changes:

• Added return_grad for all rnnt losses.
• Lifted T >= S requirement for regular case. That included making sure that shapes match not only for the whole tensor but for batch instances as well (testing boundaries if provided).
• Added s_range adjustment as per D.Povey suggestion and added unit test to check for correctness.
• Nitpicks - torch.Tensor -> Tensor, making bonudaries optional in appropriate cases.

[pkufool]

@durson I have added the github actions to this repo, could you sync the PR with the latest master, thanks!

[durson]

Merged latest master to fork and tests are passing.

[pkufool]

Thanks! merging! I would be very nice if you can also make a PR to https://github.com/k2-fsa/k2 (k2/python/k2/rnnt_loss.py).