Add is_floating_point and div_ PyTorch ops

[TylerADavis]

Thanks for contributing to TVM! Please refer to guideline https://tvm.apache.org/docs/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews from Reviewers by @ them in the pull request thread.

This PR adds support for the is_floating_point and div_ PyTorch operations, and closes #6239 .

This PR does not include a unit test for div_ as there is currently no unit test for div, and they share an implementation. Additionally, while I wrote a unit test for is_floating_point() (see https://github.com/apache/tvm/compare/main...TylerADavis:tyler_add_ops_incl_tests?expand=1), it is not included in this PR as models consisting solely of is_floating_point are compiled to constant graphs with no inputs, and the current testing infrastructure assumes that all graphs have inputs. The unit test for is_floating_point() tests the operator with torch.jit.script and torch.jit.trace with inputs consisting of int8, float64, float32, and float16. Each of these test cases passes.

[masahi]

Thanks, please run black to fix formatting.

[TylerADavis]

Hi, I just ran black on pytorch.py and pushed the changes.

[masahi]

Have you looked at verify_script_model in pytorch/test_forward.py? It might help for adding tests for is_floating_point. I also encountered an empty graph when writing a test for torch.numel with tracing. See test_numel() for its usage.

[masahi]

Thanks @TylerADavis