How to improve GROUP BY performance with CTE in Postgres

Question

I have the following simplified schema in Postgres12 whereby a user has access to many cfiles and every cfile can be accessed by many users.<