This patch allow kube-ovn-controller combine multus-cni to allocate addresses for subnet providers other than ovn by annotate pods with allocated addresses. Users need to create subnet CRs and set the provider with <name>.<namespace> of NetworkAttachmentDefinition defined by multus-cni. Then kube-ovn-controller will combine the `k8s.v1.cni.cncf.io/networks` annotation and address information in subnet cr to determine an address.

This patch only implement the controller side logical, we need other patches to complete the cni part that communicates to apiserver and fetch the address information to the main cni.

fix: only delete pod that restart policy is Always

perf: only enqueue updatePod when needed

fix: add iptables to accept container traffic

feat: check kube-proxy and coredns in diagnose

feat: add label param in install script

perf: recycle ip and lsp for pod that in failed or succeeded phase

fix: add inactivity_probe back

feat: check if crds exist in diagnose

fix: gc static routes

fix: still delete lsp if pod not in ipam

fix: delete chassis from sb when delete node

fix: missing label selector

feat: add one script installer

fix: cleanup in offline environment

refactor: the ipam now has lock itself no need for ippool queue

feat: diagnose check ds/deployment status

fix: if pod is evicted, recycle address

fix: use uuid to fetch vip

refactor ipam

Use kube-ovn ipam to replace ovn dynamic address allocation. We will design more flexible ip allocation strategy and provide ipam capacity to other cni plugin later.