Fix flakyness in tests related to torch.unique with dim=None

[SkafteNicki]

What does this PR do?

One of test seems to be flaky. See for example this CI run:
https://github.com/Lightning-AI/torchmetrics/actions/runs/10037067157/job/27736179139?pr=2636#step:16:839
Trying to debug why this is happening

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?

Make sure you had fun coding 🙃

---

📚 Documentation preview 📚: https://torchmetrics--2650.org.readthedocs.build/en/2650/

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 41%. Comparing base (ef22554) to head (c5fedd2).
Report is 107 commits behind head on master.

❗ There is a different number of reports uploaded between BASE (ef22554) and HEAD (c5fedd2). Click for more details.

HEAD has 152 uploads less than BASE

Flag	BASE (ef22554)	HEAD (c5fedd2)
Windows	8	3
python3.8	8	3
cpu	58	20
torch1.13.1+cpu	9	3
macOS	11	4
python3.10	15	5
torch2.0.1	6	2
torch2.0.1+cpu	9	3
torch1.12.1+cpu	3	1
Linux	39	13
python3.9	27	9
torch1.10.2+cpu	3	1
torch1.11.0+cpu	3	1
python3.11	8	3
torch2.4.0	3	1
torch2.3.1+cpu	6	2
torch2.1.2+cpu	3	1
torch2.2.2+cpu	6	2
torch2.4.0+cu121	3	1
torch2.4.0+cpu	2	1
torch1.13.1	2	1

Additional details and impacted files

@@           Coverage Diff            @@
##           master   #2650     +/-   ##
========================================
- Coverage      69%     41%    -27%     
========================================
  Files         316     316             
  Lines       17745   17745             
========================================
- Hits        12185    7307   -4878     
- Misses       5560   10438   +4878     

[Borda]

seems like added retry helped but it would still be worse to know what is happening there...

[SkafteNicki]

@Borda should be good now, found that the issue had to do with how torch.unique deals with multidim tensors

[Borda]

Let's remane the PR to fit the fix...