feedv2: DB-ify

Updating / reading 20K files in 1.6K folders and each time decoding and encoding the whole file  might be inefficient at some point (!). We should look into transitioning to a DB-based (data warehouse / columnar) solution.

The hope is that the barrier was strong enough to ease this transition.

Should we do this before or after cloud deployment?