Resolve prod configuration

[dolsysmith]

There's not one error I'm seeing, but a pattern of odd behavior:

• When switching from docker-compose-dev.yml to docker-compose-prod.yml, or vice versa, the Rails startup script doesn't seem to create the correct database. Even after deleting the Docker volumes and the Docker images (for rails and worker), the previous setting seems (somehow!) to persist, and it's necessary tom manually create and migrate the database in the Rails container (bundle exec rails db:create && bundle exec rails db:migrate).
• I don't think this issue is connected, but when running the prod setup, the Bulkrax import doesn't work as intended in at least one case: when a work has multiple files to be attached. I get an ldp error about a locked resource being unavailable (which is related to Fedora access). I initially thought this was a problem with the Fedora prod setup, but I don't think that's the case. (I don't get the error if I use docker-compose-dev.yml but override the fedora service to use the production settings for the latter.)

[dolsysmith]

I think know what is causing the second error above when running the app in production mode. The development setting in config/sidekiq.yml creates 1 worker; the prod setting currently creates 2. When ingesting a work with multiple filesets via Bulkrax, I suspect it's the case that the AttachFileSet jobs (one per fileset) end up on separate workers, which then try to access the same work (Fedora object) for update concurrently, with the result that one of the workers gets locked out.