dd'ing to the target device, and then growing the fs to the full disk extent seems to be a splendid idea. Especially as Linux tends to replace quite a lot of code over the usual curse of updates...

For a cluster system I had built a netbootable system for reinstall that would restore a dump of the reference system's fs on a node and adjust hostnames, ssh host keys etc... Also was quite fast at that time..