(chore): convert DeprecationWarnings to FutureWarnings where possible

[ilan-gold]

I'm not sure about PendingDeprecationWarning (which remain untouched - if you look at the codebase, there are some strange uses of them which I think is a separate issue) but otherwise this PR removes DeprecationWarning. I'm marking as 0.12 so as not to break people's CI. In that case I would probably put concrete timelines on this like "Next minor release i.e., 0.13" we will remove these.

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

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 83.67%. Comparing base (cea85f6) to head (2e873af).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1874      +/-   ##
==========================================
- Coverage   86.11%   83.67%   -2.44%     
==========================================
  Files          40       40              
  Lines        6242     6242              
==========================================
- Hits         5375     5223     -152     
- Misses        867     1019     +152     

Files with missing lines	Coverage Δ
src/anndata/_core/aligned_mapping.py	93.41% <100.00%> (ø)
src/anndata/_core/anndata.py	82.61% <ø> (ø)
src/anndata/_io/utils.py	76.15% <ø> (ø)
src/anndata/_settings.py	93.52% <100.00%> (ø)
src/anndata/utils.py	83.40% <ø> (-0.90%)	⬇️

... and 7 files with indirect coverage changes

[ilan-gold]

Maybe we remove this altogether and always force FutureWarning?

[flying-sheep]

works for me!

[flying-sheep]

PendingDeprecationWarnings are even weaker than DeprecationWarnings: They are completely ignored by default. So I say we should upgrade these too.

[ilan-gold]

PendingDeprecationWarnings are even weaker than DeprecationWarnings: They are completely ignored by default. So I say we should upgrade these too.

Yes I guess my point is that this is basically misused. One use is a parent class for OldFormatWarning where we will not remove the ability to read old files, so there is no intention to deprecate. The other place is __setitem__ for SparseDataset for which I don't have enough context right now to upgrade IMO. I would need to investigate why we are removing and what the implications are. So I want to deal with this separately.