drop_monitor: remove quadratic behavior

drop_monitor is using an unique list on which all netdevices in
the host have an element, regardless of their netns.

This scales poorly, not only at device unregister time (what I
caught during my netns dismantle stress tests), but also at packet
processing time whenever trace_napi_poll_hit() is called.

If the intent was to avoid adding one pointer in 'struct net_device'
then surely we prefer O(1) behavior.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/netdevice.h

@@ -2236,7 +2236,9 @@ struct net_device {
 #if IS_ENABLED(CONFIG_MRP)
 	struct mrp_port __rcu	*mrp_port;
 #endif
-
+#if IS_ENABLED(CONFIG_NET_DROP_MONITOR)
+	struct dm_hw_stat_delta __rcu *dm_private;
+#endif
 	struct device		dev;
 	const struct attribute_group *sysfs_groups[4];
 	const struct attribute_group *sysfs_rx_queue_group;