Migrate semantics for DBRefEntry version string to support versioned sources and cross-references

Sequence record versioning has become a pervasive issue across ENA, Ensembl, Uniprot, RNACentral, et al. Jalview has largely ignored the problem of versioning (or reconciliation of diferent versions), and whilst the DBRefEntry.version string exists in its datamodel, it was primarily used to record the origin fo the DBRefEntry (e.g. a Uniprot record yielded a version string starting with "UNIPROT.." for all DBRefEntry objects constructed from the records accessions and cross-references to other databases).

The semantics were weakened with the release of 2.11.2.6 and 2.11.3.0 - JAL-2353 became critical to fix because Uniprot records began to include fully versioned cross-references to ENA and Ensembl, causing the cross-reference retrieval mechanism to fail. The pragmatic patch enacted [~gmungoc]'s proposed change in semantics for the DBRefEntry.version field (JAL-3077), but this had (thankfully minor) side-effects (see JAL-4116).

Jalview now requires a nuanced and more complete specification of how sequence data and cross-reference versioning is managed.

* How should Jalview store versions for sequences and cross-references abstracted from public database records ?
* What should happen when a Jalview project based on older semantics is imported ?
* What options should Jalview offer when changes in version are detected ?