Apertium language pairs in Content Translation mostly gets outdated quickly, we need solution that quickly updates language pairs:
a. Fetch Apertium language pairs data from upstream repositories,
b. Put data into WMF Git repo (probably as submodule),
c. Either build automatic packages from the fetched language pairs data,
c. Put apertium language pairs data directly into 'mode_path' to use in Content Translation.
This should be approved by TechOps to avoid security/other issues.