(chore): ensure views of anndata produce distinguishable view classes

[ilan-gold]

• Closes #
• Tests added
• Release note added (or unnecessary)

[ilan-gold]

@flying-sheep Happy to add a protocol here for having base if you think that's the way to go. I am a bit ambivalent - this solution is perhaps less robust but requires exposing less API surface area.

Without API change, can go in a patch as well

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.33%. Comparing base (2e3ab38) to head (45b5300).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1637      +/-   ##
==========================================
- Coverage   86.81%   84.33%   -2.49%     
==========================================
  Files          37       37              
  Lines        6008     6008              
==========================================
- Hits         5216     5067     -149     
- Misses        792      941     +149     

Files with missing lines	Coverage Δ
src/anndata/_core/views.py	85.33% <100.00%> (-5.78%)	⬇️

... and 7 files with indirect coverage changes

[flying-sheep]

Hmm, good question.

I think if possible I’d prefer exporting _SetItemMixin under a different name. Maybe something containing COW (copy-on-write) to clarify semantics?

But it might not be possible to do that since python/typing#213 isn’t implemented yet. We might want to add our use case to https://github.com/CarliJoy/intersection_examples/issues