isolated label scores do not return expected results

[danielStrobl]

scib/scib/metrics/isolated_labels.py

Lines 229 to 230 in 82444ef

	adata.obs[iso_label_key] = adata.obs[label_key] == isolated_label
	score = silhouette(adata, iso_label_key, embed)

This function gives the mean over all cells and not only the mean for the cells that are part of the isolated label.
This leads to much lower scores with the baselines as observed here: openproblems-bio/openproblems#685

[mumichae]

should be solved by #370