Tweak Metrics collection traces

[luqmana]

The nexus log can get a bit noisy with all the successful requests to the metric collection endpoints every 10s:

Apr 15 17:27:58.715 INFO request completed, response_code: 200, uri: /metrics/collect/e6bff1ff-24fb-49dc-a54e-c6a350cd4d6c, method: GET, req_id: 4134bd35-62ea-490f-a28f-67bb6413c3ee, remote_addr: [fd00:1de::6]:47965, local_addr: [fd00:1de::7]:12221, component: dropshot_internal, name: e6bff1ff-24fb-49dc-a54e-c6a350cd4d6c
Apr 15 17:28:08.713 INFO request completed, response_code: 200, uri: /metrics/collect/e6bff1ff-24fb-49dc-a54e-c6a350cd4d6c, method: GET, req_id: 9f5530f3-81e7-489c-bb62-c2ae8afa2a8d, remote_addr: [fd00:1de::6]:47965, local_addr: [fd00:1de::7]:12221, component: dropshot_internal, name: e6bff1ff-24fb-49dc-a54e-c6a350cd4d6c
Apr 15 17:28:18.710 INFO request completed, response_code: 200, uri: /metrics/collect/e6bff1ff-24fb-49dc-a54e-c6a350cd4d6c, method: GET, req_id: 1273060b-7ca2-4d6a-bce1-c52c2062f066, remote_addr: [fd00:1de::6]:47965, local_addr: [fd00:1de::7]:12221, component: dropshot_internal, name: e6bff1ff-24fb-49dc-a54e-c6a350cd4d6c

It'd be nice to just show them on errors or a different log level perhaps.

[smklein]

I think this particular line is coming from dropshot: https://github.com/oxidecomputer/dropshot/blob/da09c39441e2d4c984d076f90cd550d9c6f76951/dropshot/src/server.rs#L722-L724

If we only log errors, I think this will impact all dropshot servers, not just Nexus. That's definitely an option - I also think being able to filter our logs would help a lot, since "under some, but not all conditions" this information could be useful.

[bnaecker]

It's true this is the normal Dropshot request handling log. It's also true that we could make these messages in particular less frequent, by running a separate server just for handling the metrics requests. There's already the ProducerServer type for this, so it would just be a matter of moving the handler functions to that server and figuring out a new address and/or port.