feat: support csc in dask arrays in get.aggregate

[ilan-gold]

• Closes sc.get.aggregate with CSC dask #3861
• Tests included or not required because:

• Release notes not necessary because:

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 76.77%. Comparing base (dd6e6fd) to head (ffad853).
⚠️ Report is 1 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3872   +/-   ##
=======================================
  Coverage   76.76%   76.77%           
=======================================
  Files         115      115           
  Lines       12365    12370    +5     
=======================================
+ Hits         9492     9497    +5     
  Misses       2873     2873           

Flag	Coverage Δ
hatch-test.pre	76.77% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/scanpy/get/_aggregated.py	92.78% <100.00%> (+0.17%)	⬆️

[flying-sheep]

As usual, just some stylistic stuff. Great work!

I’m not super happy about the added complexity, but I also don’t know if we could do it better:

e.g. it doesn’t look like we can pull out everything depending on {un,}chunked_axis and pass that in from a function that just exists to handle that complexity, things are too interwoven. What do you think?

[flying-sheep]

what do you say to what I wrote in #3872 (review)?

[ilan-gold]

what do you say to what I wrote in #3872 (review)?

Ah yeah, I meant to reply to that. I can look into it. Let's leave this PR as-is. I'll open a new one into this one fiddling around with it. If it looks to be too complex, I may just punt and say "FAU when we can"