Caching

We should have some internal data tree to cache all of the responses for INDEX requests so that if a user makes that call and then a GET/Show request that we already have, we just return that.

This would need to have some kind of expiration tracking and invalidation marking for when we make a PUT/POST request. Also will need to make available a `force` option and to update the internal data when we make DELETE requests.

This should be able to be turned off, and maybe disabled by default?

The issue names INDEX, GET/Show, PUT/POST, and DELETE request paths; start by tracing those request flows in the shard. Define the cache’s expiration, invalidation, force option, update behavior, and opt-in or default setting, then verify each request type.

Written by the indexing model from the issue text.