My goal is to create a replica of an existing database that would allow me to create new tables and populate them. I don't care about writes to replicated tables - those can and probably should throw errors.
Ideally I would like to accomplish this with standard postgresql streaming replication, but since that's not possible (or is there a way to do it?) I can go with a battle-tested and maintained extension. The least desirable solution would be a custom ETL process.
The replicated database will be used for analytics not real-time operations so synchronization delay is acceptable.
This seems like a fairly obvious use-case but I've been having trouble with finding any solution to this problem.