[Backup] Ensure the restore job can load the data in order

We need to adjust the Spark backup and restore jobs to be sure that we are loading the data in sorted order. Not only is this useful for performance today, but we may want to support some server side performance optimizations in the future that depend on this.