Export a new metric to describe each broker's state

[0xdiba]

Who is this for and what problem do they have today?

Redpanda operators don't have a metrics view of the state of each broker (whether it is active, draining, inactive, etc)
That way the cluster state as a whole cannot always be observed and monitored (eg how can operators know that no more than X brokers have gone into a decomissioning state at once).

What are the success criteria?

Having a metric that describes state of each broker (possibly reported by the controller leader only at each point in time)

Why is solving this problem impactful?

Gives easier operational insights for the cluster and its brokers

JIRA Link: CORE-4259