This will hopefully detect potential LVM (or any other storage, when
they implement it) issues before committing changes just on some
nodes.

Unfortunately due to the dry_run opcode handling, we can't integrate
this into the usual handling (as we need to activate the disks before
doing any tests, which belongs in Exec not in CheckPrereq).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>