Initial metrics support

[lukesteensen]

We'd like the router to support both transforming logs into metrics and collecting metrics directly. For the first pass at this, we'll probably need a few basic components:

• The ability to represent metrics in our internal data format. Both cernan and SSF could be good inspiration here.
• A transform that can aggregate its input and emit output on a scheduled interval
• A simple way to send that output somewhere we can see it. This could be as simple as treating it as log lines we can use with existing sinks, or just logging it out.

Some things to think about but not necessarily solve right away:

• How "typed" do we want our internal data format? There's a whole spectrum between a simple map of attributes and full fledged "log line" / "metric" types with that apply to transforms/sources/sinks
• How do will we support "bucketing" of aggregated metrics? Purely based off wall clock time of their arrival, or will there be a concept of event time vs processing time? How would late data be handled in that case?

[binarylogic]

Noting, the goal for this first version is something simple that works. More of an experimental exercise than a full complete version. I'd recommend we start accepting metrics from a specific type (ex: statsd) and then simply forwarding that data to the same system. That will at least ensure we support the metrics data type in a fundamental way.