Fix and add gff getRegionFeatureDensity calculations

[cmdcolin]

This attempts to make the autogenerated gff3tabix histogram a little more accurate by incrementing all histogram bins covered by features from state to end (previously it only incremented first bin)

This also adds getRegionFeatureDensities to the regular GFF3 adapter

I also removed one of the sample tracks because it seemed confusing that it used a different feature density bigwig that didn't accurately reflect underlying data

Screenshot

[cmdcolin]

Would fix #913

[rbuels]

@cmdcolin could you add this to the release notes?