Take a look at the script responsible for generating the backup here http://svn.linuxmce.com/pluto/branches/charon-merge/src/BootScripts/PlutoRestorePoint.sh
From A brief glance at it, it appears to backup
-files needed by the web admin
-the pluto.conf config file
Basically, it would be used to backup your installation so that you can restore that backup on a clean install if you had to do a reinstall. It doesn't backup media or other files not specified.
Regarding running a RAID on the root - I would advise against it. There was a recent bug report filed in mantis where someone tried this very thing. LMCE found the superblock and auto-detected the RAID and treated it as it would any other RAID array - causing cyclic sysmbolic links and other problems. This may be fixed in a future version, but for now I wouldn't do it.
Also - regarding RAID. They aren't bulletproof. I have lost data and so have others on a RAID array. It helps, but it also helps to understand how it (mdadm) works underneath so you know what to do when there is a problem. Asside from that, I don't think running the OS on a RAID is needed anyways. The OS is free and replaceable. The data isn't. If I were you I would:
1) Use a separate system disk for the OS
2) Create the backup points in the web admin on a schedule. If your OS ever did go bad, you can clean install a new one and restore from the backup (this should take care of most things except custom changes to config files)
3) Keep your data on a RAID array. This should give you some redundancy as well as the ability to do a reinstall on your system disk without fear of overwriting your data. Also, check the RAID status in the web admin at least weekly, as LMCE currently does not alert you on a disk failure. On top of that, a hot spare drive in your array is also a good idea