Add support for protein sequences

[fethalen]

This PR addresses issue #93 by introducing support for amino acid sequences.

1. qiime alignment mafft can now accept either FeatureData[Sequence] or FeatureData[ProteinSequence] as an input, the resulting output will be of the type FeatureData[AlignedSequence] or FeatureData[AlignedProteinSequence] depending on the input type.
2. qiime alignment mafft-add works in a similar manner, but takes an alignment as an additional output. The sequence type of the alignments and the sequences to be added must match.
3. Additionally, qiime alignment mask can now accept either FeatureData[AlignedSequence] OR FeatureData[AlignedProteinSequence] as an input.

[cherman2]

Let's do a larger typemap here. https://develop.qiime2.org/en/latest/plugins/references/api/types.html#qiime2.plugin.TypeMap

I think this will look something like:

T_alignment, T_sequence, T_expanded_alignment = (FeatureData[AlignedSequence], FeatureData[AlignedSequence]) : FeatureData[AlignedSequence]

I think with this typemap, you can remove _validate_sequence_pair() helper.

[cherman2]

Hi @fethalen,
Thanks for all your work on this. We have some ideas for how to simplify this code. Also, since we merged your alignment strategy PR, there are merge conflicts for this PR.

Let us know if you have any questions or need anything!

[fethalen]

I've redefined the dependent types as suggested and was able to remove the _validate_sequence_pair() helper function. In addition, I've simplified the way that the sequence type is inferred, and have removed the SequenceType enum. A lot of the tests have been removed as well, since this were related either to the helper function or the enum. I don't see a way moving forward using two directory formats as the output of the _mafft helper function.

[cherman2]

Makes sense regarding the union not being a valid output. I didn't think about that. Your original code makes sense given that limitation.

Let's Get This Merged!