don't uncordon masters

upgrade-cluster.yml

@@ -67,7 +67,6 @@
     - { role: kubernetes/node, tags: node }
     - { role: kubernetes/master, tags: master }
     - { role: network_plugin, tags: network }
-    - { role: upgrade/post-upgrade, tags: post-upgrade }
 
 #Finally handle worker upgrades, based on given batch size
 - hosts: kube-node:!kube-master