`cBioPortal` datasource annotates some genomic mutations incorrectly

[mtiberti]

In particular, it looks like the assembly for the mutation is incorrect in some caes.

This is because the cBioPortal data source is hardcoded to save genomic mutations as hg19, as that was the standard when it was first developed. More recently, cBioPortal switched to a mix of hg19 and hg38, making it so that we store some mutations encoded in hg38 as hg19