Test performance of reversible vs irreversible hafah schemas using 2 postgresTs
I have a suspicion based on some preliminary performance analysis that hivemind (and probably other apps) would massive sync faster if it used views that only referenced the irreversible tables (avoiding the join to reversible tables). Fortunately, this theory is easily tested: we can temporarily redefine the views that hivemind's massive sync uses to be just the irreversible tables and compare the speed again a normal massive sync. Start with a 5M massive sync, then try a 20M massive sync. Hivemind is particularly interesting as this is our slowest syncing app right now.