(fix): correct default fill values for dask-sparse

[ilan-gold]

• Noticed in (feat): read_lazy for whole AnnData lazy-loading + xarray reading + read_elem_as_dask -> read_elem_lazy #1247
• Tests added
• Release note added (or unnecessary)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.45%. Comparing base (097377c) to head (ebb173d).
Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1719      +/-   ##
==========================================
- Coverage   86.93%   84.45%   -2.49%     
==========================================
  Files          40       40              
  Lines        6039     6039              
==========================================
- Hits         5250     5100     -150     
- Misses        789      939     +150     

Files with missing lines	Coverage Δ
src/anndata/_core/merge.py	83.91% <100.00%> (-11.08%)	⬇️

... and 7 files with indirect coverage changes

[flying-sheep]

Doesn’t look like we use the test “is a sparse matrix/array or a Dask array containing a sparse matrix/array” anywhere else so I’m not suggesting to make this a helper in this PR yet

[flying-sheep]

Dask sparse is only in 0.11 right? I modified the milestones under that assumption. If I’m wrong and you want to backport this, please change it back

[ilan-gold]

I don't think so, just the API to do i/o is

[flying-sheep]

OK, needs manual backport then