changes available_suttas.py into a "catalog manger" which generates available_suttas.json (which is used by index.js) and files_to_cache.json which is now used by sw.js. This means we will not need to manually change sw.js anytime a new sutta is added. Both files are updated in the repo by github actions. We can see if this step becomes too fragile and breaks the workflow. The current tests should catch any outright breaking of the site.
sw.js has some addition changes that should also make it be able to download new suttas without clearing the cache (pending further testing)
closes #27
This PR: