Use alignments from tRNAscan-SE without re-creating them in R2DT

[patriciaplchan]

Instead of creating tRNA alignments with cmalign and tRNA covariance models, the alignments generated by tRNAscan-SE in the secondary structure output can be used. The alignments generated by tRNAscan-SE use the domain-specific models, not isotype-specific models. The exceptions are mitochondrial tRNAs in vertebrates and selenocysteine tRNAs that use isotype-specific models. tRNAscan-SE also makes adjustments to the alignments originally generated by cmsearch. These include removing the non-Watson-Crick basepairs, adding missing basepair that sometimes occurs at the two ends of tRNA-fMet in bacteria, and removing extra helix such as variable arm or D/T-arm. That means when using tRNAscan-SE alignments, the SS and SS_cons included in the .afa file as input for infernal2mapping.py will always be the same. Since the current templates were generated from the consensus of the model training data, I expect that they will mostly match up with the secondary structure of individual tRNAs. The main variation will be the length of the variable arm for Type II cytosolic tRNAs, and the length of the D-arm and T-arm of the mitochondrial tRNAs. If Infernal/Traveler mapping can handle this variation, I think the current templates should just work. But I can certainly make adjustments to the templates if needed after trying out the fixes of the Infernal/Traveler mapping problem in combination with the use of tRNAscan-SE alignments.