Add a /health API endpoint

[corviday]

We are using Prometheus, which can scrape arbitrary pages to collect performance metrics, so any performance data the backend makes available via a /health API endpoint could be incorporated into our monitoring and logging systems for the live servers.

Cache hit/miss/size data would be useful for catching or diagnosing problems. We still have a frustratingly hard to reproduce and intermittent memory issue that realtime insight into the cache could conceivably shed light on, and in general, information on cache efficiency would help fine-tuning load sharing between the backend instances.