Question: How to speed up min/max aggregates in Postgres without an index that is unnecessary otherwise