Since we want to support many databases, in some cases them pointing
to the same primary DB. The `database_tasks:` allows to model
that whether the migration tasks should be skipped (create, drop, migrate)
depending on how the data is being stored.