Avoid a restart loop with NM dispatcher scripts

Newer NetworkManager versions will run the dispatcher scripts when new
unknown interfaces show up, including 'clat'. That will cause a clatd
restart right after startup, and we'll never get to fully initialise. So
ensure the dispatcher script does nothing if the device in question is
our own 'clat' to prevent this.

scripts/clatd.networkmanager

@@ -7,6 +7,13 @@
 # Written by Tore Anderson <tore@fud.no>
 #
 
+# Newer NetworkManager versions will run the dispatcher scripts once
+# a new unmanaged interface shows up, including the 'clat' interface
+# created by clatd/TAYGA. So if we're being called due to our own
+# interface showing up, do nothing, otherwise we will end up
+# committing suicide from the restarts below
+[ "$DEVICE_IFACE" = "clat" ] && exit 0
+
 # We simply restart clatd in all situations, as no matter if an interface
 # goes up or down, it may mean that the PLAT device changes, it may mean
 # native IPv4 appearing or disappearing, or it may mean that DNS64 became