Patch 6 deserves some explanation, as does patch 7.
For preexisting devices, if there is encryption below the layers
we can alter (eg: lvm pvs) we should never ever set the encryption
checkbutton sensitive since it is impossible to turn off encryption
for such a device without removing it. We already had this covered
when setting up the customization area, but I dropped the ball in
the on_reformat_toggled handler by not taking this complexity into
account.
Patch 7 adds a udev_settle() call at the bottom of udev_trigger.
This prevents the possibility of getting into DeviceTree.populate
before the udev queue has settled from our trigger call, which could
potentially lead to missing info in the udevdb.