Hmm, alas no, not exactly.
The LEM archive of the normalized data will basically replicate any data that isn't already on the archive store, and it'll grow forever (more or less, as long as it can push that archive data off). But, it's a mirror/copy, not a move and delete sort of operation. If you used that model, you would leave your LEM appliance at one storage level, then use the archive for slower/long term/wasteful storage. If you had to pull something back in, it gets a little complicated, we'd need to figure out what to pull back in and probably do it on a secondary system to not affect/confuse the running system. (We're looking at ways to improve that bit, either making it easier to have a secondary "I just need to search these dates for a little while out of my backup" or use the remote archive directly as frozen storage).