Update annotation schema for strand changes

[victorlin]

Description of proposed changes

Related issue(s)

Follow-up to #1211 and #966 (comment).

Testing

• Add new tests

Checklist

• Add a message in CHANGES.md summarizing the changes in this PR that are end user focused. Keep headers and formatting consistent with the rest of the file.

[jameshadfield]

Conversation moved from #1281 (comment):

Entries in the annotations JSON (except "nuc") correspond to translated CDSs, and are interpreted by Auspice as such. In this context, it makes no sense to be strandless / unknown-strand. As of nextstrain/auspice#1684, Auspice will skip any entry if the strand is not "+"/"-".

If we want to extend our concept of annotations to include features beyond CDSs then it makes sense to include these, although I'd prefer we use GFF syntax for the enum values. However I'm not sure what we'd do with these features -- we can't translate them, and Auspice can't (currently) display them. One potential example of us doing this would be nextstrain/auspice#1686, which I think in GFF terms would be represented as a strandless feature, but I think we'll encode it as a sub-feature of the CDS.