remove out of core median

[Intron7]

Computing counts per cell for normalize_total leads to better performance.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 76.49%. Comparing base (ac19bb3) to head (7d91abc).
Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3379      +/-   ##
==========================================
- Coverage   76.58%   76.49%   -0.09%     
==========================================
  Files         111      111              
  Lines       12862    12865       +3     
==========================================
- Hits         9850     9841       -9     
- Misses       3012     3024      +12     

Files with missing lines	Coverage Δ
src/scanpy/preprocessing/_normalization.py	85.36% <100.00%> (-0.35%)	⬇️

... and 8 files with indirect coverage changes

---- 🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[flying-sheep]

Fair enough. Can this blow up? I.e. is counts ever big enough that this is a problem?