Add _LieGroupCheckContext setting for disabling internal warnings

[jacoblubecki]

Motivation and Context

Normalization in Lie group checks can be overly verbose in logging, it was identified that it could be an improvement if there was a way to selectively disable these checks in certain contexts (e.g. when implicit normalization is desired).

#486

How Has This Been Tested

Existing unit tests + manual testing in python REPL.

Example REPL test:

>>> non_se3 = torch.Tensor([[100, 0, 0, 0], [100, 100, 0, 0], [0, 0, 100, 0]]).unsqueeze(0)
>>> with enable_checks(), set_lie_group_check_enabled(True):
        th.SE3(tensor=non_se3)

~/Projects/theseus/theseus/theseus/geometry/manifold.py:141: UserWarning: The input tensor is not valid for SE3 and has been normalized.
  warnings.warn(
SE3(tensor=tensor([[[ 0.8944, -0.4472,  0.0000,  0.0000],
         [ 0.4472,  0.8944,  0.0000,  0.0000],
         [ 0.0000,  0.0000,  1.0000,  0.0000]]]), name=SE3__1)
>>> with enable_checks(), set_lie_group_check_enabled(True, silence_internal_warnings=True):
        th.SE3(tensor=non_se3)

SE3(tensor=tensor([[[ 0.8944, -0.4472,  0.0000,  0.0000],
         [ 0.4472,  0.8944,  0.0000,  0.0000],
         [ 0.0000,  0.0000,  1.0000,  0.0000]]]), name=SE3__2)

Types of changes

• Docs change / refactoring / dependency upgrade
• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

Checklist

• My code follows the code style of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have read the CONTRIBUTING document.
• I have completed my CLA (see CONTRIBUTING)
• I have added tests to cover my changes (no existing tests for this specific context manager, and code is covered in other existing tests)
• All new and existing tests passed (***except CUDA tests which I could not run locally)

[jacoblubecki]

I thought using the @contextmanager decorator might help simplify the code a little bit -- I believe the functionality should be the same since no internal code was using any of the subclasses directly, and since this API is internal, downstream code shouldn't break, but if you'd prefer I leave the class-based approach, I am happy to revert back to that.

[luisenp]

Hi @jacoblubecki, thanks for adding this! Sorry for the delay, I had this in my TODO stack but somehow two weeks passed!

I haven't tested the code yet, but the code and the example lgtm. What we want now is to turn on this new option under the hood when running an optimizer, so that intermediate Lie group operations don't throw this warning if there are numerical precision errors that are not necessarily the users' fault. The best place to do this should be this line. Let me know if you have any questions. x