draining stuck on hard instance failure

[svenwltr]

The node-drainer gets stuck on a hard instance failure. In this case:

• the node gets NotReady in Kubernetes
• the pods on this node get either Unknown or NodeLost
• the instance is in Terminating:Wait state

It gets stuck, because:

1. The node gets not removed from Kubernetes, because it is still in the ASG.
2. The instance does not get removed from the ASG, because it waits for the draining.
3. The draining cannot happen, because the node is actually gone.

Also the node is nor reachable via SSH nor ping.