it seems if you deploy one using agentPoolProfile it fills out the defaultNodePool in the outputs automatically, then on the second pass it thinks it needs to destroy the cluster to remove it