Minimum contig length for alternate code prediction

[snayfach]

A feature request to specify the minimum contig length for which prodigal-gv should try and predict a non-standard genetic code. Suggest using 10kb as the default. I believe this was your suggestion from our discussion, but wanted to create an issue to track it.

I looked at the rate in which prodigal-gv predicts alternatives codes in IMG/VR data. In large contigs > 20kb, prodigal-gv predicts alternative codes for ~1.5% of viral contigs. This increases to 2.5% <10kb, 3.3% <5kb, and 5.4% <2.5kb. My hunch is that most of the alternative code predictions for short contigs are FPs.