Might make sense to run the reindex against pg_attribute off peak hours

I'll test in db-lab today and write up the CR if that shows improvement for the bloat query

This ticket was created from INC-3404 and was automatically exported by incident.io 🔥