Prefetch and cache languages.json instead of per request

We can assume that the languages.json file is nearly constant, or updated at most once per week.

Let's not fetch it per request.

Locate the request path that currently fetches languages.json and inspect how its response is used. Confirm the current per-request behavior, then verify that the file is prefetched and reused rather than fetched for every request, with updates no more often than the stated weekly assumption.

Written by the indexing model from the issue text.