Showing deleted nodes

eks-node-viewer seems to be showing nodes that aren't there. I suspect they used to exist and they have since been deleted.

```
x.compute.internal cpu ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (0 pods) r7i.8xlarge/$2.1168 On-Demand - NotReady/93d
x.compute.internal cpu ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (0 pods) r7i.24xlarge/$6.3504 On-Demand - NotReady/92d
x.compute.internal cpu ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (0 pods) r7i.16xlarge/$4.2336 On-Demand - NotReady/90d
x.compute.internal cpu ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (0 pods) r7i.16xlarge/$4.2336 On-Demand - NotReady/83d
x.compute.internal cpu ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (0 pods) r7i.24xlarge/$6.3504 On-Demand - NotReady/75d
x.compute.internal cpu ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (0 pods) r7i.2xlarge/$0.5292 On-Demand - NotReady/75d
x.compute.internal cpu ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0% (0 pods) r7i.4xlarge/$1.0584 On-Demand - NotReady/68d
```

When I run a `kubectl get nodes` it doesn't show these dead nodes. I checked AWS to see if the terminated nodes are listed. It's difficult to tell because the dns hostname is gone so I can't match exactly which one is which, but `eks-node-viewer` shows way more dead nodes than the ec2 console shows terminated instances.

I'm sure there is a dead record somewhere, but I don't know where to clean it up.

No source file or test is named. Start by reproducing the discrepancy between eks-node-viewer and `kubectl get nodes`, then trace the node-discovery path and compare its results with the Kubernetes node list. Done means deleted or absent nodes are no longer displayed, with a regression test covering the stale-node case.

Written by the indexing model from the issue text.