Fix index types

[flying-sheep]

Ilan made some good progress here, so this is intended to finish things up.

We have two “deal with indices” helpers in anndata._core: .index._normalize_index which accepts broad user-supplied types and unifies them into simple versions, and .views._resolve_idxs which simplifies multiple successive indexing operations (views of views).

Currently we have several “1D” (along the obs or var axis) index types which I’ll refer to by their name in the code:

1. Index1D: to capture user-supplied values like list[str], int, pd.Series[bool] and so on.
2. Index1DNorm (new): stored in anndata views as _oidx/_vidx: Can be a slice or np.ndarray with dtype np.bool_ or np.integer
3. what we pass to _init_as_actual: can be Index1DNorm or an int/np.integer. We should probably get rid of this and sliceify 0D indices in _normalize_index

TODO:

• check if all advertised types work and clarify (e.g. which pd.Index[???] do we support?)
• check if subsetting views works in all permutations (e.g. mask → slice or mask → ndarray)

[flying-sheep]

I don’t think we test these (at least not via subset_func and therefore with test_views_of_views)

[codecov]

Codecov Report

❌ Patch coverage is 91.07143% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.55%. Comparing base (7ff7e3b) to head (66d90f1).
⚠️ Report is 49 commits behind head on main.

Files with missing lines	Patch %	Lines
src/anndata/tests/helpers.py	81.81%	4 Missing ⚠️
src/anndata/_core/views.py	87.50%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2067      +/-   ##
==========================================
- Coverage   86.81%   85.55%   -1.26%     
==========================================
  Files          46       46              
  Lines        7052     7067      +15     
==========================================
- Hits         6122     6046      -76     
- Misses        930     1021      +91     

Files with missing lines	Coverage Δ
src/anndata/_core/anndata.py	83.07% <100.00%> (+0.17%)	⬆️
src/anndata/_core/index.py	92.65% <100.00%> (ø)
src/anndata/_core/raw.py	75.53% <100.00%> (-4.32%)	⬇️
src/anndata/_core/xarray.py	98.39% <100.00%> (+21.39%)	⬆️
src/anndata/compat/__init__.py	80.90% <100.00%> (+0.17%)	⬆️
src/anndata/_core/views.py	85.83% <87.50%> (-5.31%)	⬇️
src/anndata/tests/helpers.py	82.72% <81.81%> (-10.83%)	⬇️

... and 6 files with indirect coverage changes