select/filter reads matching/not-matching regions of interest

We can currently count() reads overlapping with regions defined in a GFF file.
However this is only accessible for counting.

Another use-case is within select() in order to keep/drop reads that overlap a feature set.
This is useful to reduce a SAM/BAM file to regions or features of interest.

mapped = select(mapped) using |mr|:
    mr = mr.filter(min_match_size=45, min_identity_pc=90, action={match})
    mr = mr.intersect(action={match}, gff_file="regions_of_interest.gff",
                      features=["chr"], subfeatures=["highGC"])
    if mr.flag({mapped}):
        discard

A less flexible alternative but easier since we already bundle samtools, would be exposing samtools view to work with a target region.

Start by tracing the existing count() support for GFF-defined regions and the select() filtering flow, then compare the proposed intersect() and samtools view approaches. Done should mean select() can keep or discard SAM/BAM reads based on overlap with the requested GFF features or target regions.