In the install documentation for EKS, it has you manually remove some stale AWS iptables rules. Not removing these rules results in subtle and random connectivity issues, most notably around host-network processes accessing the pod network. This typically results in a failure to reach the health check.

So, when we're sure these aren't in use, we should remove these in the nodeinit script.

The challenge is knowing these rules are not in use. From reading the install documentation, I don't see a reliable way to detect it, so we'll have to add another helm value. Probably something like nodeinit.removeAWSCNI.