Compatibility layer for pandas3 changing fill_value for sparse from 0 to NaN

[timtreis]

pandas 3.0 changed DataFrame.sparse.from_spmatrix to use fill_value=NaN instead of fill_value=0 for float columns (pandas PR #59064). With fill_value=NaN, groupby().mean() excludes those zeros from the denominator, inflating means and producing wrong results in ligrec.

The fix wraps the from_spmatrix call in PermutationTestABC.__init__ with a small helper that resets any fill_value=NaN sparse columns back to fill_value=0. It's skipped on pandas <3.

[codecov]

Codecov Report

❌ Patch coverage is 66.66667% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 66.21%. Comparing base (3c11223) to head (2ecac93).

Files with missing lines	Patch %	Lines
src/squidpy/gr/_utils.py	60.00%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1118      +/-   ##
==========================================
- Coverage   66.21%   66.21%   -0.01%     
==========================================
  Files          44       44              
  Lines        7163     7168       +5     
  Branches     1217     1218       +1     
==========================================
+ Hits         4743     4746       +3     
- Misses       1943     1944       +1     
- Partials      477      478       +1     

Files with missing lines	Coverage Δ
src/squidpy/gr/_ligrec.py	77.28% <100.00%> (ø)
src/squidpy/gr/_utils.py	60.34% <60.00%> (-0.02%)	⬇️

🚀 New features to boost your workflow:

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

[selmanozleyen]

hi I have an alternative solution to this. Normally something simpler is enough for this PR but I also wanted to get rid of an unreliable function so I created this PR: #1119

[timtreis]

Closed in favor of #1119