Add some metrics from the production system to show how stable the design works

[sttomm]

To back up the slide in the presentation that shows how easy it is to operate all of this in production, it would be great if some metrics could be shown, like how often do messages get into the retry-ladder, etc.