Consider breaking out the collapsing of similar sequences from `digestFastqs`

[csoneson]

The main reason would be that currently there is no guarantee that different samples are collapsed identically.

[csoneson]

Downstream effects/modifications related to this:

• With the collapsing out of digestFastqs(), consider making it possible to process multiple variable sequences in different ways in there - provide a list of parameters and a list of WT sequences, each as long as the number of variable regions (separately for fwd and rev, if applicable). Loop over the matching to WT part; if the corresponding list element is empty (e.g. for a variable sequence that is a barcode), just keep the sequence. Otherwise, match with the corresponding WT sequences.
• Mutant naming - consider using different string separators for "within variable region", "between variable regions of the same (fwd/rev) reads", and "between fwd and rev" for easier parsing.

Other considerations:

• A permitlist of barcodes?
• Add a count threshold for inclusion in the tree
• Set min counts for collapsing to be at least equal to the count ratio (assuming all counts are natural numbers)