Add documentation to stop Unicorn/Puma and Sidekiq pods before restoring from backup
As described in gitlab-org/gitlab#36405 (comment 342151060) and gitlab-org/gitlab!40789 (merged), having processes that can write to GitLab tables while restoring from a database backup can cause duplicate rows be inserted and schema changes to fail quietly. At the moment, we don't have a good story of restoring a database backup in a single transaction, so we need to tell admins to stop Puma and Sidekiq.
In https://docs.gitlab.com/ee/raketasks/backup_restore.html#restore-for-omnibus-gitlab-installations, we mentioned stopping running processes, but we don't mention any sort of thing in https://docs.gitlab.com/13.2/charts/backup-restore/restore.html. Since https://gitlab.com/gitlab-org/build/CNG/blob/master/gitlab-task-runner/scripts/bin/backup-utility#L171 appears to use the same gitlab-backup restore
task, I think we need to document this.