Just making here for placeholder for potentially something that could be worth looking into
James Bonfield at one point told me record counter, which is currently used for creating a unique ID for jbrowse does not actually need to be unique in the file. I have not thus far seen issues with it but that might not be the rule
defined in cram PDF ashttps://github.com/samtools/hts-specs/blob/master/CRAMv3.pdf
ltf8 record counter 1-based sequential index of records in the file/stream