* Split response size into normal responses and AXFR|IXFR responses.
* Split out incoming sizes as well in normal and AXFR|IXFRs.
* Add meta qtype to the monitored qtypes.
* Make duration to be exported in milliseconds instead of seconds.
Make middleware that sets up a (http) handler survive a graceful
restart. We calls the middleware's Shutdown function(s). If restart
fails the Start function is called again.
* middleware/health: OK
* middleware/pprof: OK
* middleware/metrics: OK
All restart OK.
Add prometheus metrics to the cache handler. This just used prometheus,
if the metrics middleware does not setup the handler, there is nobody
reading these metrics, but they are still reported. Seems the simplest
solution while keeping the whole middleware separation in tact.
