More robust metric state comparison

[SkafteNicki]

What does this PR do?

Fixes bug found in discussion #1016
When searching for compute groups in metric collection we need to compare the states between metrics. torch.allclose is not robust to difference in dtype such as np.allclose (see issue pytorch/pytorch#55356). PR makes this comparison more robust.

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure to update the docs?
• Did you write any new necessary tests?

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

[codecov]

Codecov Report

Merging #1022 (b3eece8) into master (6cffeb5) will decrease coverage by 0%.
The diff coverage is 86%.

@@          Coverage Diff           @@
##           master   #1022   +/-   ##
======================================
- Coverage      95%     95%   -0%     
======================================
  Files         180     180           
  Lines        7824    7828    +4     
======================================
  Hits         7431    7431           
- Misses        393     397    +4