Migration Job sometimes hangs when updating services

[dlbrittain]

When running update_services_oneoff.sh the migration job that checks to see if the database needs to migrate the metadata table models sometimes hangs until the timeout (600s default) is reached. This job is launched if we have configured any replicas to be deployed that interact with the database (the Annotation Engine, Materialization Engine or any Celery Worker replicas). A better model might be to check what is being deployed to avoid launching the migration job if the service in question is not directly leveraging database models that might change. This will primarily improve the update_services_oneoff.sh script.