clusterd: Dereference CLP tcp socket after hangup

We have a small leak here, that clp sockets keep referenced after the
peer closed. Fix.

clusterd/clusterd

@@ -1056,7 +1056,7 @@ sub clp_receive {
 	my $data;
 	defined $s->recv($data,$TCP_MAX) or return;
 	if (!length($data) ) {
-		delete $mgmt_sockets{$s};
+		delete $clp_sockets{$s};
 		$s->close;
 		return;
 	}