Better handling of mRNAs with multiple CDSs

[standage]

Technically, an mRNA feature with multiple coding sequences is not invalid. In fact, the canonical gene from the GFF3 spec includes such a case (alternative start codons). In practice, different CDSs often have their own dedicated mRNA features, even if those mRNAs have identical structure.

Recently I've come across a couple of cases in NCBI GFF3 files where mRNAs have multiple CDSs, but in each case it appeared to be an erroneous duplicated annotation. Diagnostic messages from AEGeAn weren't that helpful: basically stating that the following assertion had failed in agn_typecheck_feature_combined_length.

agn_assert(strcmp(id, fid) == 0);

This particular issue should probably be handled somewhere upstream, so that mRNAs with multiple CDSs are flagged before this function is called on them. Whether to issue a warning and proceed or to halt at once needs to be considered.