netkit: Allow for configuring needed_{head,tail}room

Allow the user to configure needed_{head,tail}room for both netkit
devices. The idea is similar to 163e529200af ("veth: implement
ndo_set_rx_headroom") with the difference that the two parameters
can be specified upon device creation. By default the current behavior
stays as is which is needed_{head,tail}room is 0.

In case of Cilium, for example, the netkit devices are not enslaved
into a bridge or openvswitch device (rather, BPF-based redirection
is used out of tcx), and as such these parameters are not propagated
into the Pod's netns via peer device.

Given Cilium can run in vxlan/geneve tunneling mode (needed_headroom)
and/or be used in combination with WireGuard (needed_{head,tail}room),
allow the Cilium CNI plugin to specify these two upon netkit device
creation.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Acked-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://lore.kernel.org/bpf/20241220234658.490686-1-daniel@iogearbox.net
Signed-off-by: Ihor Solodrai <ihor.solodrai@linux.dev>

include/uapi/linux/if_link.h

@@ -1315,6 +1315,8 @@ enum {
 	IFLA_NETKIT_MODE,
 	IFLA_NETKIT_SCRUB,
 	IFLA_NETKIT_PEER_SCRUB,
+	IFLA_NETKIT_HEADROOM,
+	IFLA_NETKIT_TAILROOM,
 	__IFLA_NETKIT_MAX,
 };
 #define IFLA_NETKIT_MAX	(__IFLA_NETKIT_MAX - 1)