Proposal to change how kops performs validation during rolling-updates

[dskatz]

/kind feature

The kops validation checks that are performed during rolling-update procedures tend to timeout on kubernetes clusters that are dynamically sized frequently. The dynamic nature of the work causes kops validation to fail because those nodes entering or leaving will essentially be in a NotReady state.

This request is to start the conversation on potential changes to how kops performs validation checks or if there is a different approach managing dynamic clusters with kops.