Dashboard state for external resources can get out of sync for terminated resolutions

[augray]

When you terminate a resolution, it only has a certain amount of time to clean up resources before the pod is force-killed by k8s. This might not be enough time to detect that the resource has in fact been cleaned. The resolver is the one that causes the server to actually check resource state against the external entities it represents. Thus when the resolver dies before detecting cleanup is done, it can leave the dashboard saying the resource is permanently in the "deactivating" state.

This appears to have happened here