Sum up statistics on insertion length at a specific position in snpcov

[cmdcolin]

the idea of summing up the data about insertions was sort of alluded to by @ihh when we were thinking about the bilby encoder

could also be done for skip and deletion and maybe other types also. clipping length is not as relevant as it is just sort of the ragged end of a read but it could be done.

precedence: summing up statistics on all reads in a column was implemented by modification probabilities (ML tag) recently also for methylation

[cmdcolin]

note that this can also be taken to a "extreme" level and the snpcov could, in principle, sum up lots of weird things like
-percent of reads that have a particular tag at this position (for e.g. HP tag)
-percent of reads with particular BAM flag
etc.