varlens allele support is too slow

[timodonnell]

Running on anything more than a few thousand variants is painful. I think @armish gave up trying to do this with varlens since he has tens of thousands of variants. This could be improved.

[armish]

Just to provide more context to this: my specific task was to extract number of reads covering a particular region of the genome and I had to perform this task thousands of times. Here is a really simple benchmarking result:

$ time varlens-reads --reads test/data/CELSR1/bams/bam_5.bam --locus chr22:46930257 |wc -l > /dev/null

real    0m5.191s
user    0m1.416s
sys 0m0.213s

$ time samtools view test/data/CELSR1/bams/bam_5.bam chr22:46930257-46930257 |wc -l > /dev/null

real    0m0.090s
user    0m0.021s
sys 0m0.003s

Though we should note that varlens is an over-kill for this particular need as it comes with way more features than needed for this laser-focused task ;)