Make a template that demonstrates instance monitoring and restarting.
We use cfn-push-stats to send a heartbeat and if we don't get
it within the set interval we restart the instance.
Other changes to make this work:
- clear the waitcondition when it is deleted.
- set the resource id to None when deleted, else it won't re-create
properly.
- don't run watch alarms if the stack is not completed.