chore: disable zarr v2 writing (at least as far as we can control it)

[ilan-gold]

2 reasons to do this:

1. Fewer tests mean faster completion
2. No one should be writing v2 data anymore given that all reasonable data is in-spec writable and spatial data has upgraded (and anndataR can now read v3 data as well: https://anndatar.scverse.org/reference/write_zarr.html?q=v3#zarr-format)

I'm setting this as a 0.13.4 to remind myself to backport the functionality by deprecating instead of removing the setting. I will then add a test for that as well as a sensible release note for the 0.13.4 release.

• Closes #
• Tests added

• Release note not necessary because:

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.80%. Comparing base (d75540e) to head (f30a494).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2628      +/-   ##
==========================================
- Coverage   87.66%   85.80%   -1.86%     
==========================================
  Files          49       49              
  Lines        8121     8114       -7     
==========================================
- Hits         7119     6962     -157     
- Misses       1002     1152     +150     

Files with missing lines	Coverage Δ
src/anndata/_io/zarr.py	86.25% <ø> (-0.31%)	⬇️
src/anndata/_settings.py	100.00% <100.00%> (ø)

... and 8 files with indirect coverage changes