Get capi targetsize from cache

This ensured that access to replicas during scale down operations were never stale by accessing the API server kubernetes#3104.
This honoured that behaviour while moving to unstructured client kubernetes#3312.
This regressed that behaviour while trying to reduce the API server load kubernetes#4443.
This put back the never stale replicas behaviour at the cost of loading back the API server kubernetes#4634.

This PR tries to satisfy both non stale replicas during scale down and prevent the API server from being overloaded. To achieve that it lets targetSize which is called on every autoscaling cluster state loop from come from cache.

Also note that the scale down implementation has changed https://github.com/kubernetes/autoscaler/commits/master/cluster-autoscaler/core/scaledown.

cluster-autoscaler/cloudprovider/clusterapi/clusterapi_nodegroup.go

@@ -18,6 +18,7 @@ package clusterapi
 
 import (
 	"fmt"
+	"github.com/pkg/errors"
 
 	corev1 "k8s.io/api/core/v1"
 	"k8s.io/apimachinery/pkg/apis/meta/v1/unstructured"
@@ -52,7 +53,14 @@ func (ng *nodegroup) MaxSize() int {
 // (new nodes finish startup and registration or removed nodes are
 // deleted completely). Implementation required.
 func (ng *nodegroup) TargetSize() (int, error) {
-	return ng.scalableResource.Replicas()
+	replicas, found, err := unstructured.NestedInt64(ng.scalableResource.unstructured.Object, "spec", "replicas")
+	if err != nil {
+		return 0, errors.Wrap(err, "error getting replica count")
+	}
+	if !found {
+		replicas = 0
+	}
+	return int(replicas), nil
 }
 
 // IncreaseSize increases the size of the node group. To delete a node