Update isolated_labels.py

[danielStrobl]

As mentioned in #367, the isolated label metrics do not compute what they should.
By calculating the individual silhouette scores and then taking the mean over only the cells of the cluster, this should be fixed.

[LuckyMD]

I guess you'll need to adapt the tests as well for this.

[LuckyMD]

Conceptually this looks correct, if silhouette_samples gives you the silhouette score per cell. See comment above about adding tests.

[codecov]

Codecov Report

Merging #368 (3aa8303) into main (1be0daa) will not change coverage.
The diff coverage is 100.00%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #368   +/-   ##
=======================================
  Coverage   53.13%   53.13%           
=======================================
  Files          22       22           
  Lines        1848     1848           
=======================================
  Hits          982      982           
  Misses        866      866           

Flag	Coverage Δ
unittests	53.13% <100.00%> (ø)

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

Impacted Files	Coverage Δ
scib/metrics/isolated_labels.py	94.33% <100.00%> (ø)

[LuckyMD]

Please also adapt the perfect score test

[danielStrobl]

ah bit too fast

[LuckyMD]

@mumichae could you still review this please?