App Deployment after a failed app deployment requires manual cleanup

Bug Description

A metadata and Postgres schism seems to have occurred during a failed app deployment. I now have to manually query postgres to find enums, objects, and fields that need to be deleted to clear the road for the next app deployment. Recover:Reinstall does not find metadata records for the remaining entities and so cleanup is left to a developer or dba.

Expected behavior

I wish I had a command from twenty that could look at the manifest, more deeply compare what objects would be created and if they already existed, and optionally perform the update skipping those existing entities or deleting them first. something like yarn twenty:command deepClean --try-leave-in-place.