we should abstract some of the genomic location info up to a more generic class that can be used for gene etc. Also to allow the same variant to have positions on >1 assembly
exonId seems overly specific. If we want to model intersection with and effect on genomic entities then consider a more sophisticated model. Presumably cardinality > 1 here?
Merging in as a starting point, but let's follow-up on https://github.com/phenopackets/phenopacket-format/issues/10 this may be refactored
Particularly