Refactor the DB to make the schema better

eg. we really don't separate admin nd main collections, with mostly redundent info, do we? Just a variable that says paper unapproved or approved might be enough

No files, tests, or entry points are named. Start by locating the database schema and mapping the admin and main collections, including their redundant fields; the work is done only after the intended schema and approval-state representation are clearly agreed.

Written by the indexing model from the issue text.