in-process sqlite is great for simple DB-agnosticc stuff, but sometimes you want to point to a PG server
Would like things like:
- point to a pg db to be used, probably clearing it out first
- (maybe) point to a pg DB cluster and have it do the whole "create a random DB, do the thing, then cleanup"... maybe
- (double maybe) point to PG binaries and then just spin up a cluster to be thrown away???