Split merged fastqs and merge aligned bams after markduplicates

[anoronh4]

Currently the pipeline creates one result per input line in the mapping file. We currently do not split fastqs that were previously merged, and we do not accurately assign read groups in final alignment file. It is possible that we will perform picard's MarkDuplicates (this is useful for QC and certain downstream analysis tools)

This issue can be split into two parts:

1. splitting input fastqs before any other steps
2. merging fastqs (not all tasks require proper RGIDs in alignment) or merging bams.