Add the size (on disk) of the backup files to the stats files

cbbackupmgrs info command outputs the aggregated size of the backup files on disk, this is extremely useful but could become quite costly when interrogating an archive stored in AWS S3. We already avoid having to get the item/mutation/deletion count from S3 by writing stats files upon completion of a backup; we should add the sizes of the backups files (on disk) to these stats files. This will mean when running info against an archive stored in the cloud we can avoid having to make 2048 (Rift meta/data store) API calls per bucket.