i32 limits

[ekg]

Great library! My only concern is that it seems to be hard coded to 32-bit signed integers for ranges and queries. There are a lot of genomes where there are chromosomes that exceed these limits. Either we special case these or relax and use 64 bits. How would you like a patch to enable 64 bit types?

[dcjones]

The SIMD versions have to be specialized on 32 bit integers, but the generic implementation could be pretty easily made generic on coordinate type. It would be just be a matter of adding another type parameter to the interval types, traits, and BasicCOITree type. This would be a worthwhile change, but my (maybe very human and mouse centric) impression was few work with chromosomes larger than 2gb so it hasn't been a priority.