Add-concat-on-disk-examples

[flying-sheep]

• Closes “How to” example for concat_on_disk #1160
• Tests added (automatic doctest)
• Release note added (or unnecessary)

REST API docs: https://api.cellxgene.cziscience.com/curation/ui/#/

Uses a simpler version of https://gist.github.com/ivirshup/f1a1603db69de3888eacb4bdb6a9317a (maybe they also changed the schema since then)

[codecov]

Codecov Report

Merging #1161 (9bf0b30) into main (afc5930) will increase coverage by 0.22%.
The diff coverage is 84.61%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1161      +/-   ##
==========================================
+ Coverage   82.58%   82.81%   +0.22%     
==========================================
  Files          36       36              
  Lines        5364     5364              
==========================================
+ Hits         4430     4442      +12     
+ Misses        934      922      -12     

Files	Coverage Δ
anndata/_core/anndata.py	83.47% <ø> (ø)
anndata/experimental/merge.py	91.28% <84.61%> (+3.58%)	⬆️

... and 2 files with indirect coverage changes

[flying-sheep]

@selmanozleyen We don’t have the overwrite argument in other commands. Is there a specific reason why you included it?

[selmanozleyen]

@selmanozleyen We don’t have the overwrite argument in other commands. Is there a specific reason why you included it?

@flying-sheep
Ignore my previous message please. Sorry I confused the branches. I just wanted it to have that option as only taking filenames seemed too restrictive. But in theory the user can give zarr/h5py objects as input too so it shouldn't be a problem.

[ivirshup]

Made some minor changes:

• concat_on_disk now errors if there are no objects to concatenate (like ad.concat, pd.concat, xr.concat, np.concatenate)
• concat_on_disk now errors if out_file's parent directory does not exist (like anndata.write_h5ad, pd.DataFrame.to_csv)