docs(auroc): clarify per-batch logit/probability check in BinaryAUROC

[mohansree14]

What does this PR do?

Fixes #2195

BinaryAUROC.update() (inherited from BinaryPrecisionRecallCurve) checks per batch whether preds looks like logits (values outside [0, 1]) and applies sigmoid if so. This check runs independently on every update() call rather than once across the full accumulated dataset, so with small batch sizes it's possible for some batches to be treated as probabilities and others as logits within the same run, producing inconsistent results.

Per the reporter's suggested resolution, this PR documents that behavior more prominently in BinaryAUROC's class docstring rather than changing the underlying per-batch logic (which other metrics also rely on), to keep this a minimal, low-risk documentation fix. Happy to extend this to the base BinaryPrecisionRecallCurve docstring too, or take a different approach, if maintainers prefer.

Before submitting

• Was this discussed/agreed via a GitHub issue? (Conversion from logits to probabilities happens on a batch by batch basis #2195)
• Did you read the contributor guideline?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes? (n/a — this PR is the doc update)
• Did you write any new necessary tests? (n/a — documentation-only change, no behavior change)
• Did you verify new and existing tests pass locally?
• Did you list all the breaking changes introduced by this pull request?

This is my first contribution to this repo — feedback welcome.

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 31%. Comparing base (d184220) to head (91b9915).
⚠️ Report is 12 commits behind head on master.
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

❗ There is a different number of reports uploaded between BASE (d184220) and HEAD (91b9915). Click for more details.

HEAD has 784 uploads less than BASE

Flag	BASE (d184220)	HEAD (91b9915)
torch2.0.1+cpu	20	2
python3.10	129	13
Windows	30	3
cpu	218	22
torch2.9.1	9	1
macOS	39	4
python3.12	89	9
torch2.9.1+cpu	30	3
torch2.8.0+cpu	30	3
torch2.8.0	10	1
Linux	149	15
torch2.10.0	10	1
torch2.7.1+cpu	20	2
torch2.10.0+cpu	20	2
torch2.4.1+cpu	10	1
torch2.2.2+cpu	10	1
torch2.1.2+cpu	10	1
torch2.5.1+cpu	10	1
torch2.0.1	10	1
torch2.3.1+cpu	9	1
torch2.6.0+cpu	10	1

Additional details and impacted files

@@           Coverage Diff            @@
##           master   #3474     +/-   ##
========================================
- Coverage      37%     31%     -6%     
========================================
  Files         349     349             
  Lines       19901   19910      +9     
========================================
- Hits         7264    6170   -1094     
- Misses      12637   13740   +1103     

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.