fix: resolve test-suite warnings

[Zethson]

Summary

Cleans up the warnings emitted while running the test suite. Running pytest on main produced 329 warnings; with this change the runnable subset produces 0. No library behavior changes.

Fixed at the source

• Augur logistic regression no longer passes the penalty argument that scikit-learn 1.8 deprecated (FutureWarning, removal in 1.10). The L1/L2 mix is now expressed with l1_ratio and the unregularized case with C:
  • l2 → l1_ratio=0.0 (verified to produce identical coefficients to the old default, default lbfgs solver, works across the supported sklearn range)
  • l1 → l1_ratio=1.0, elasticnet → l1_ratio=0.5 (both with the saga solver)
  • none/None → C=np.inf
• statsmodels DGE test sets the negative-binomial alpha=1.0 explicitly — the previous implicit default — removing the ValueWarning raised when the parametrization is collected.
• Perturbation-space test wraps the intentional ensure_consistency=False add/subtract calls in pytest.warns(UserWarning, ...), which both silences the leaked warning and asserts the warning is actually raised.

Filtered as unavoidable third-party noise

Added to the existing filterwarnings list in pyproject.toml (same pattern already used for ~15 other dependency warnings):

• pydeseq2 "dispersion trend curve fitting did not converge" and "residual degrees of freedom is less than 3" — emitted on the small synthetic data the Milo and compare_groups tests use (this was ~310 of the 329 warnings).
• jaxopt "no longer maintained" deprecation notice.
• joblib loky "A worker stopped while some jobs were given to the executor" notice.
• formulaic DataMismatchWarning from no-intercept contrast handling in the DGE base tests.

Notes

• The sklearn penalty deprecation was fixed in code rather than filtered, since it originates from pertpy's own call.
• 10 failures / 106 errors seen locally are pre-existing network-blocked dataset downloads in this sandbox and are unrelated to this change.

🤖 Generated with Claude Code

[codecov-commenter]

Codecov Report

❌ Patch coverage is 80.00000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 77.80%. Comparing base (5fa8ed7) to head (113563d).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
pertpy/tools/_augur.py	80.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1018      +/-   ##
==========================================
- Coverage   77.81%   77.80%   -0.01%     
==========================================
  Files          50       50              
  Lines        6580     6584       +4     
==========================================
+ Hits         5120     5123       +3     
- Misses       1460     1461       +1     

Files with missing lines	Coverage Δ
pertpy/tools/_augur.py	89.64% <80.00%> (-0.22%)	⬇️

🚀 New features to boost your workflow:

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