Proposal: Add Drain Precheck

[svenwltr]

We could add some prechecks before draining a node.

An example would be a check whether there are any pending pods in the cluster. This would reduce the damage if some admin is too eager with scaling down the instances. Example scenario:

• admin needs to replace one ASG with another
• the new ASG is ready and the old ones are cordoned
• the admin scales down the old ASG in chunks of 3
• the new ASG actually is not big enough, therefore scheduling pod from the last chunk might fail
• node-drainer happily continues draining the remaining nodes