fix slow calculations of classification metrics

[Isalia20]

What does this PR do?

Fixes #2868

Before submitting

• Was this discussed/agreed 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? Yes

Make sure you had fun coding 🙃

Details

The flag for the _TORCH_GREATER_EQUAL_1_12 was removed on the changed line in the following commit: 346bcdc
Now problem with the removal is that if before condition for checking was:
if torch.are_deterministic_algorithms_enabled() or _XLA_AVAILABLE or _TORCH_GREATER_EQUAL_1_12 and x.is_mps:

which meant if tensor was on mps and torch was greater or equal 1.12 it would choose the conditional path. With removal of the flag _TORCH_GREATER_EQUAL_1_12, now for the code to enter this path the XLA should be available and tensor should be on mps and I don't think there is such a device at all that supports both XLA and mps tensors. After changing the code the performance issue on the issue posted above gets fixed

---

📚 Documentation preview 📚: https://torchmetrics--2876.org.readthedocs.build/en/2876/

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 69%. Comparing base (4d9c843) to head (7279d82).
Report is 3 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #2876    +/-   ##
=======================================
- Coverage      69%     69%    -0%     
=======================================
  Files         346     332    -14     
  Lines       19142   18966   -176     
=======================================
- Hits        13230   13055   -175     
+ Misses       5912    5911     -1