Clean stale background migration jobs

In !24135 (merged), we reverted the problematic ActivatePrometheusServicesForSharedClusterApplications background migration, but the migration only dropped jobs that were in the background_migration Sidekiq queue. There may be more jobs in the Sidekiq scheduled or retry sets, so we need to clear those too.

Closes https://gitlab.com/gitlab-com/gl-infra/infrastructure/issues/9160