Add read-side cache for stats and metrics to centralize and thereby reduce I/O during requests

Instead of reading these two files from disk whenever a request related to them is serviced, we just read in both every 60 seconds and keep one copy in memory all the time and service the requests based on those. This should significantly reduce our disk I/O as soon as at least more than one request every 60 seconds is served for any of the affected routes.

Closes #225 (closed)