Create a *Metrics event for every role

One of the most commonly used events that we write to trace logs are our *Metrics events, of which we have several. For example:

`ProcessMetrics`
`NetworkMetrics`
`MemoryMetrics`
`StorageServerMetrics`
`TLogMetrics`
`CommitLatencyMetrics`

Some of these events are logged for specific roles (e.g. `StorageServerMetrics`) and provide a lot of utility for understanding what that role is doing.

We have some roles, however, that we don't these events. The ones that come immediately to mind are ratekeeper and data distributor, which instead rely on a variety of other less standardized events to piece together what's happening (e.g. `RkUpdate`, `MovingData`, etc.).

I think it would be helpful to have a `*Metrics` event available for each role that contains the data that might commonly need to be known by an operator of the cluster. Data in these metrics could also be candidates for pushing through status, if they aren't there already.