Add Syriac to Reader

[anlinguist]

This PR adds Syriac Peshitta to the Reader Generator. Unfortunately, the Syriac TF data doesn't contain the gloss attribute, so there's not yet a way to add a vocabulary.

Also, I've modified the Dockerfile a bit to make rebuilding the container faster by separating the setup stuff from the file data so that when file data is changed, the container can use cached layers for setup and only load in the changed files.

[anlinguist]

[anlinguist]

I would have liked to make each text configurable for Hebrew vs Syriac, to allow having texts of both languages in the same PDF, maybe a future PR.