docs(pp.scale): document that variance/std use ddof=1 (sample correction)

[AxelMcKenna]

• Closes #
• Tests included or not required because: docs-only change, no behavior is added or modified.

• Release notes not necessary because:

---

sc.pp.scale computes gene-wise variance and standard deviation with Bessel's correction (ddof=1), via mean_var(x, axis=0, correction=1) in scale_array and scale_array_masked. This convention isn't stated anywhere in the docstring, so a reimplementation using numpy.std() defaults (ddof=0, population variance) silently diverges from sc.pp.scale's output - negligible for large n_obs, but non-negligible for small datasets (e.g. ~0.02% on a ~2700-cell PBMC3k-sized dataset, enough to shift where max_value clipping takes effect).

This PR adds a .. note:: to the scale docstring stating the correction used. No behavior change.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 79.85%. Comparing base (ecec3c2) to head (4a7ba69).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4196   +/-   ##
=======================================
  Coverage   79.85%   79.85%           
=======================================
  Files         121      121           
  Lines       12924    12924           
=======================================
  Hits        10321    10321           
  Misses       2603     2603           

Flag	Coverage Δ
hatch-test.low-vers	78.97% <ø> (ø)
hatch-test.pre	79.78% <ø> (-0.03%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/scanpy/preprocessing/_scale.py	89.58% <ø> (ø)