Infere dns_service_ip from service_cluster_ip_range

when setting a custom service_cluster_ip_range label, the default cluster dns service ip (10.254.0.10) may end being outside the service IP range which results in DNS not working in the cluster.

Heat templates already support setting the needed variable. Automatically take the 10th ip address from the service range. This should fix the issue of DNS not working for people who set the service_cluster_ip_range label while keeping things unchanged for those who don't set it.

magnum/drivers/common/templates/kubernetes/helm/cern-chart.sh

@@ -138,8 +138,14 @@ ${NVIDIA_GPU_VALUES}
       enabled: ${MONITORING_ENABLED}
       grafana:
         adminPassword: ${GRAFANA_ADMIN_PASSWD}
+      kubeControllerManager:
+        endpoints: ${KUBE_MASTERS_PRIVATE}
       kubeEtcd:
         endpoints: ${KUBE_MASTERS_PRIVATE}
+      kubeScheduler:
+        endpoints: ${KUBE_MASTERS_PRIVATE}
+      kubeProxy:
+        endpoints: ${KUBE_NODES_PRIVATE}
       prometheus:
         prometheusSpec:
           externalLabels: