ovs, dpdk: fix creating ovs-interface when the ovs-bridge is netdev

When the ovs-bridge datapath is netdev, OpenvSwitch will not create a
ovs-interface but a tun interface. The ovs-interface device must check
all the link-change signals and check if the link type is tun and the
interface name is the same than the device name. If so, the
ovs-interface device will get the ifindex of the tun device. This allow
NetworkManager to manage the interface properly, modifying MTU,
configuring IPv4/IPv6 and others.

Example:

```
55: ovsbridge-port0: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 9000 qdisc fq_codel state UNKNOWN group default qlen 1000
    link/ether fa:fb:07:98:e0:c6 brd ff:ff:ff:ff:ff:ff
    inet 192.168.123.100/24 brd 192.168.123.255 scope global noprefixroute ovsbridge-port0
       valid_lft forever preferred_lft forever
    inet6 fe80::9805:55c4:4c5f:da1c/64 scope link noprefixroute
       valid_lft forever preferred_lft forever
```

https://bugzilla.redhat.com/show_bug.cgi?id=2001792
NetworkManager/NetworkManager!1179
54 jobs for fix_ovs_ip in 31 minutes and 40 seconds (queued for 5 seconds)
latest