Once pg_sphere is installed and we agreed on obscore configurations (DM-36701) we need to run migrations to add everyhting to each of the repos.

My initial idea is that migration script would just add obscore config to butler_attributes and create the obscore teable with a correct schema (which is based on config). For actual filling of the table from pre-existing data I want to implement a separate script which will run later, that way the migration can be short and we don't need a downtime for it. The population script can run concurrently with other users, so we should not need any disruption.

Just like for dimensions manager we'll have two separate migrations branches, one for the manager class itself, and another for configuration. It is also likely that we'll need separate migration branch for each repo, as each of them will heave an independent configuration. IT makes sense to start with a single repo as a demonstrator to show people how it's going to look like. It is also less critical for obscore to do everything right on the first iteration, we can always delete and re-create obscore table again.