fix: chan's parallel mean-var algorithm for dask-backed arrays

[flying-sheep]

like scverse/scanpy#4143

currently all Claude, I need to go over it

[codecov]

Codecov Report

❌ Patch coverage is 98.18182% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 96.76%. Comparing base (e22249f) to head (a508509).

Files with missing lines	Patch %	Lines
src/fast_array_utils/stats/_mean_var.py	98.03%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #200      +/-   ##
==========================================
- Coverage   99.33%   96.76%   -2.57%     
==========================================
  Files          21       21              
  Lines         598      649      +51     
==========================================
+ Hits          594      628      +34     
- Misses          4       21      +17     

☔ 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-hq]

Merging this PR will degrade performance by 26.37%

⚠️ 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

❌ 16 regressed benchmarks
✅ 216 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_stats_benchmark[numpy.ndarray-2d-all-int32-mean_var]	359 ms	492.8 ms	-27.15%
❌	test_stats_benchmark[scipy.sparse.csr_array-2d-all-int32-mean_var]	359.3 ms	493.1 ms	-27.13%
❌	test_stats_benchmark[scipy.sparse.csc_array-2d-all-int32-mean_var]	359.4 ms	493.2 ms	-27.13%
❌	test_stats_benchmark[scipy.sparse.csc_array-1d-all-int32-mean_var]	359.4 ms	493.2 ms	-27.13%
❌	test_stats_benchmark[numpy.ndarray-2d-ax1-int32-mean_var]	359 ms	492.7 ms	-27.12%
❌	test_stats_benchmark[scipy.sparse.csr_array-1d-all-int32-mean_var]	359.4 ms	493.1 ms	-27.12%
❌	test_stats_benchmark[numpy.ndarray-1d-all-int32-mean_var]	359.3 ms	492.8 ms	-27.09%
❌	test_stats_benchmark[numpy.ndarray-2d-all-float32-mean_var]	364.5 ms	493.6 ms	-26.15%
❌	test_stats_benchmark[scipy.sparse.csc_array-2d-all-float32-mean_var]	364.8 ms	493.9 ms	-26.13%
❌	test_stats_benchmark[scipy.sparse.csc_array-1d-all-float32-mean_var]	364.8 ms	493.9 ms	-26.13%
❌	test_stats_benchmark[numpy.ndarray-1d-all-float32-mean_var]	364.5 ms	493.4 ms	-26.12%
❌	test_stats_benchmark[numpy.ndarray-2d-ax1-float32-mean_var]	364.5 ms	493.4 ms	-26.12%
❌	test_stats_benchmark[scipy.sparse.csr_array-2d-all-float32-mean_var]	364.9 ms	493.9 ms	-26.12%
❌	test_stats_benchmark[scipy.sparse.csr_array-1d-all-float32-mean_var]	364.9 ms	493.9 ms	-26.11%
❌	test_stats_benchmark[numpy.ndarray-2d-ax0-int32-mean_var]	401 ms	534.7 ms	-25.01%
❌	test_stats_benchmark[numpy.ndarray-2d-ax0-float32-mean_var]	406.3 ms	535.4 ms	-24.12%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing fix-dask-precision (a508509) with main (e22249f)}