mergeCells should have an option to sub-cluster a given cell type

[patrickjdanaher]

This will require mergeCells to take in the counts and neg info. This seems reasonable.

[patrickjdanaher]

Approach:

• just call nbclust, or call full insitutype?
• needed args: counts, neg, bg, cohort
• hard-code subset sizes and convergence criteria when calling the insitutype from mergeCells

Need to update name. Call it "refineClusters".

New Inputs:

• subcluster = list("a" = 2, "f" = 3:4)

[patrickjdanaher]

Subclustering logic:

• run insitutype on just the cell type in question
• replace old cluster's loglik with updated logliks

[patrickjdanaher]

implemented and merged into cohorting branch, which has not been merged into Main yet.

[patrickjdanaher]

merged