Story: Pod / server restarts should not cause cache to be lost

[markpbaggett]

Problem

Last week, we had to change an environmental variable in Rancher. Doing this caused us to lose all of the irIIIFService cache since it's only configured to use Redis and store manifests in memory. There should be a safe way to restart so that all manifests aren't loss in case of a pod restarting or a node being moved. This is especially problematic as some manifests take well over an hour to regenerate.