Problematically, older scanpy versions have the read import and we probably want to support older versions for a while since https://github.com/scverse/scanpy/tree/6d234a7ec8c33348b7778e9ccb8fe7f226e4723d is only three weeks old. So fixing this probably requires bumping the minimum supported version of scanpy with anndata, which I don't think we are ready to do.

Codecov Report

Attention: Patch coverage is 96.55172% with 1 line in your changes missing coverage. Please review.

Project coverage is 82.99%. Comparing base (af6480e) to head (8ea1c4e).

@@            Coverage Diff             @@
##             main    #1767      +/-   ##
==========================================
- Coverage   87.01%   82.99%   -4.02%     
==========================================
  Files          40       40              
  Lines        6059     6056       -3     
==========================================
- Hits         5272     5026     -246     
- Misses        787     1030     +243     

Since this isn't until 0.12 I would like to wait to see if numba comes out with python 3.13 support and then drop 3.10 and do all of the scientific python ecosystem all at once: #1768

Then we don't have to worry about register