Fix multipletests compatibility with NumPy >= 2.0

[aaronspring]

Summary

• Replaces vectorize=True with no core dims in multipletests — which called statsmodels_multipletests on each scalar p-value individually — with input_core_dims=[["s"]] so the entire flattened array is passed at once. NumPy 2.0 removed the implicit coercion of 1-element arrays to scalars in np.vectorize, causing ValueError: setting an array element with a sequence.
• Adds a _multipletests_numpy helper that broadcasts the scalar alphacSidak/alphacBonf outputs to match the shape of the p-value array (required since all 4 outputs now share output_core_dims=[["s"]])
• Fixes test_multipletests_alpha to use explicit well-controlled p-values instead of random Pearson-r p-values; with proper BH correction, random p-values around 0.19–0.93 are not rejected at any alpha, hiding the test assertion

Test plan

• pytest xskillscore/tests/test_stattests.py — 27/27 passed
• python -m pytest --doctest-modules xskillscore --ignore xskillscore/tests — 32/32 passed
• Full suite: 2794 passed, 3 skipped

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.57%. Comparing base (eccc957) to head (755f8f2).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #449      +/-   ##
==========================================
+ Coverage   94.49%   94.57%   +0.08%     
==========================================
  Files          27       27              
  Lines        2836     2841       +5     
==========================================
+ Hits         2680     2687       +7     
+ Misses        156      154       -2     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

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