when add egress qos for pod, the pod network is unreachable.
through research, the container's virtual nic's qlen is 0 is
the root reason. so set the nic's qlen to 1000(default for linux)
when create pod.

refer: https://github.com/ovn-org/ovn-kubernetes/pull/869

1. update ovs to fix ovn-nbctl daemon issues.
2. ping gw when node init to activate flow.

When reboot node, the lost interface and qos will prevent pod deletion and new pod creation.