SAMFileHeader in CRAM files is missing attributes required by the CRAM spec

[cmnbroad]

AFAICT, CRAM files created with htsjdk contain the samFileHeader that was passed in to the CRAM writer, whether or not it has the attributes required by the CRAM spec.The following are missing:

• The SQ:MD5 checksum is required unless the reference sequence has been embedded into the file.
• At least one RG record is required.
• UR (URI for the sequence fasta) field is strongly advised

I think we should fix this before we merge in #478, and add corresponding checks to the CRAMFileReader to validate these on input for >= v3.0 files.

[vadimzalunin]

ah, yes, this is done in cramtools by i missed it here...

[cmnbroad]

We really should fix this. See samtools/htslib#769.

[cmnbroad]

I don't see any reference in the spec about a read group requirement, so I assume thats been removed, and will only implement the MD5 requirement.