Add OpInfo for torch.Tensor.__setitem__ (to test prims.copy_with_setitem)

Nobody has claimed this yet.

Currently we have a manual test at https://github.com/Lightning-AI/lightning-thunder/blob/1575797a8d21e0b845e188ddd6af305be6457b1c/thunder/tests/test_ops.py#L242-L253

and PR https://github.com/Lightning-AI/lightning-thunder/pull/1322 extends this test.

I think it would be nice if we can leverage the OpInfo infra though I am not sure how it interacts for inplace operations (as currently all ops in opinfo are out-of-place).

We have some infra which uses OpInfo to generate a new list `_inplace_opinfo` for inplace ops by finding their corresponding `inplace` version. (Maybe we can directly add the OpInfo for setitem here?)

https://github.com/Lightning-AI/lightning-thunder/blob/1575797a8d21e0b845e188ddd6af305be6457b1c/thunder/tests/test_inplace_functionalization.py#L108-L109

cc @mruberry @apaz-cli

Start with the manual test in thunder/tests/test_ops.py around lines 242-253 and review PR #1322, then inspect the _inplace_opinfo handling in thunder/tests/test_inplace_functionalization.py around lines 108-109. Determine how torch.Tensor.__setitem__ should be represented in OpInfo for prims.copy_with_setitem; done means the OpInfo-based coverage replaces or extends the manual coverage and the relevant tests pass.

Written by the indexing model from the issue text.