Implement KEP-3836

TL;DR: we want to start failing the LB HC if a node is tainted with ToBeDeletedByClusterAutoscaler.
This field might need refinement, but currently is deemed our best way of understanding if
a node is about to get deleted. We want to do this only for eTP:Cluster services.

The goal is to connection draining terminating nodes

cmd/kube-proxy/app/server.go

@@ -27,6 +27,7 @@ import (
 	"strings"
 	"time"
 
+	"k8s.io/kubernetes/pkg/features"
 	utilnode "k8s.io/kubernetes/pkg/util/node"
 
 	"github.com/fsnotify/fsnotify"
@@ -869,6 +870,11 @@ func (s *ProxyServer) Run() error {
 	if s.Config.DetectLocalMode == kubeproxyconfig.LocalModeNodeCIDR {
 		nodeConfig.RegisterEventHandler(proxy.NewNodePodCIDRHandler(s.podCIDRs))
 	}
+	if utilfeature.DefaultFeatureGate.Enabled(features.KubeProxyDrainingTerminatingNodes) {
+		nodeConfig.RegisterEventHandler(&proxy.NodeEligibleHandler{
+			HealthServer: s.HealthzServer,
+		})
+	}
 	nodeConfig.RegisterEventHandler(s.Proxier)
 
 	go nodeConfig.Run(wait.NeverStop)