chore: use setup_cache for asv benchmarks

[ilan-gold]

Hopefully this speeds things up by not writing to disk for every benchmark when avoidable

• Things started looking quite slow with fix: remove read_dataset usage in read_lazy{_elem} stack #2158 and fix: fancy indexing fixes backed h5py error #2066
• Tests added
• Release note added (or unnecessary)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.68%. Comparing base (7c52d36) to head (a41178d).
⚠️ Report is 1 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2164      +/-   ##
==========================================
+ Coverage   79.84%   85.68%   +5.83%     
==========================================
  Files          46       46              
  Lines        7111     7111              
==========================================
+ Hits         5678     6093     +415     
+ Misses       1433     1018     -415     

see 19 files with indirect coverage changes

[scverse-benchmark]

Benchmark changes

Change	Before [7c52d36]	After [a41178d]	Ratio	Benchmark (Parameter)
-	18.3±0.5ms	14.5±0.3ms	0.79	sparse_dataset.SparseCSRContiguousSlice.time_getitem('first', False)

Comparison: https://github.com/scverse/anndata/compare/7c52d36628cc593c064a89c45b601bb256c8630d..a41178d2b7b852bcdef2b6a1d6c25de625878c36
Last changed: Wed, 22 Oct 2025 16:25:48 +0000

More details: https://github.com/scverse/anndata/pull/2164/checks?check_run_id=53398765209