aws: don't drain node if we can't scale in

[hsmade]

It seems the autoscaler selects and drains a node, to scale in, even if the ASG forbids it (which it finds out after it's done draining and continues to try to delete the VM).

[jrasell]

Hi @hsmade and thanks for raising this issue. It sounds like we should add a pre-flight check before draining the node to ensure the node can be terminated from the ASG. I'll mark this for roadmapping.