Dealing with multiple taxids by taking first one (iss#54)

[nawrockie]

Removes requirement that each accession only have 1 taxid from GenBank, now we take first taxid parsed from GenBank xml; this is github issue #54.

Not sure if you want to merge this now or later (or never) - up to you.

[kalvari]

It makes sense to use an umbrella tax id for each accession belonging to a specific genome assembly. This means that for simplicity purposes all accessions in the genome inherit the tax id associated with high level assembly, regardless its actual taxonomic classification.

Two main things to keep an eye on:

1. Increased redundancy: For new sequences imported from SEED alignments; If the sequence does not belong to a genome which is a member of the current non-redundant genome collection annotated in Rfamseq, overtime this will introduce redundancy and orphan sequences as seen in Rfamseq 12
2. Increased complexity for sequence export: Sequence export will need to rely on tax ids rather than genome/proteome accessions, possibly leading to redundant hits for different assembly versions of the same genome. A possible workaround is the separation between SEED and FULL hits during export.

Possible features affected:

• FTP fasta file generation (currently supported)
• Sequence indexing (release process)
• RNAcentral sequence export (release process)
• Sequence extraction https://docs.rfam.org/en/latest/sequence-extraction.html