Suggestion: Add expected results

[milesmatthias]

Before suggesting destroying the cluster, a list of things to notice about the setup, along with some basic scenarios to try, would be helpful. Ideas:

• Spiking CPU on node, seeing corresponding alerts / metrics in Stackdriver
• Stopping node, seeing K8s automatically create new node, and see corresponding events & logs in Stackdriver
• Health check setup
• Point directly to source code that does logging to Stackdriver