ci: fix bench variance

[flying-sheep]

the 100× loop doesn’t cut it, let’s do worst-case instead.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.92%. Comparing base (83974b6) to head (b1b433a).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #219      +/-   ##
==========================================
- Coverage   99.31%   97.92%   -1.39%     
==========================================
  Files          21       21              
  Lines         584      579       -5     
==========================================
- Hits          580      567      -13     
- Misses          4       12       +8     

☔ 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.

[codspeed]

Merging this PR will improve performance by ×3.3

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 7 improved benchmarks
✅ 225 untouched benchmarks

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	test_stats_benchmark[scipy.sparse.csr_array-2d-ax1-int32-is_constant]	676 ms	73.4 ms	×9.2
⚡	test_stats_benchmark[scipy.sparse.csc_array-2d-ax0-int32-is_constant]	318.7 ms	77.4 ms	×4.1
⚡	test_stats_benchmark[scipy.sparse.csc_array-2d-ax0-float32-is_constant]	319.9 ms	85.3 ms	×3.7
⚡	test_stats_benchmark[scipy.sparse.csr_array-2d-ax1-float32-is_constant]	297.1 ms	85.4 ms	×3.5
⚡	test_stats_benchmark[scipy.sparse.csc_array-2d-ax0-float64-is_constant]	309.5 ms	115.1 ms	×2.7
⚡	test_stats_benchmark[scipy.sparse.csr_array-2d-ax1-float64-is_constant]	315.9 ms	120.3 ms	×2.6
⚡	test_stats_benchmark[scipy.sparse.csr_array-2d-ax1-int32-mean_var]	99.7 ms	83.4 ms	+19.54%

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing better-slow (b1b433a) with main (83974b6)}