[Torch] Fix advanced indexing with NoneType index arguments

[padreofthegame]

Suppose to solve issue #12922.

[tvm-bot]

Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.

• cc @shingjan, @yelite _{See #10317 for details}

_{Generated by tvm-bot}

[shingjan]

Thanks for taking a look at this @padreofthegame! Will circle back once this PR is ready for review

[padreofthegame]

This solution is very similar to #13306, and represents workaround for the problem mentioned in #12922 for PyTorch models (since advanced indexing does not work the same for example for numpy and for pytorch, I put solution in torch frontend, because issue was reported for torch models).
Also, I put WIP label only because I am not sure if I managed to cover all possible test cases, so @shingjan you are welcome to try some additional scenarios, and call me out if detect something strange.

[shingjan]

@padreofthegame i think this fix is fine in terms of the difference between adv_index of numpy and it of pytorch. There is no numpy frontend for TVM yet. I think overall this PR LGTM. Will be great if @masahi can take a look and approve. Thanks!

[shingjan]

LGTM