address #159

[daler]

The original helpers.canonical_transcripts was not well tested (or even used much as far as I know).

@EFraim, this uses your PR #156 but I was unable to update it myself and fix the merge conflict (because understandably your master branch as diverged in the 2 yrs since you've addressed this!)

Eventually I would like to modularize this to instead have a method on the FeatureDB that finds the canonical transcript for a gene, and have the sequence part dealt with separately, e.g.,

for gene in genes_of_interest:
    transcript = db.canonical_transcript(gene)

but that will be later...