Prometheus metrics for sync-status

[jbearer]

Feature request from P2P: include the information from /node/sync-status in prometheus metrics. The challenge here is that prometheus metrics are inherently push-based: whenever a metric changes, some task is responsible for updating the registry. Whereas the way /node/sync-status currently works is pull-based: we only run it when queried, and it's quite expensive to run.

To make this work, we might need to develop an incremental way of tabulating this metric, so we can update Prometheus whenever it changes. We can then potentially use this incremental method to make the sync-status endpoint itself more efficient, although it should still do a full db scan at some frequency, so we catch things like corrupted or missing data, e.g. due to an RDS failover.