Group Timings Visibility

When designing an alerting pipeline, it's important for an operator to understand how the semantics of the system map their understanding.

The Alertmanager has some knobs such as:

• Group Interval (Global and routing specific)
• Repeat Interval (Global and routing specific)
• Group Wait (Global and routing specific)
• HA_peer timeout (HA mode only + per replica)

Currently, there are no metrics or visibility into any of these - so operators are left on "waiting" to see what happens or checking to see if any of the log lines make sense.

I think we can do better and provide visibility into the current state of all groups, so that operators can answer questions such as:

• What this is alert group was last evaluated?
• What state are these alert groups that I see?
• Which replica is firing what?
• What is the state of this replica?

I recognise that these answers are not simple when we think about alertmanager in HA mode and there's a chance of divergence but I think we can do better than what we currently do at the moment.

No files, tests, or entry points are named. Start by locating how group timing and HA state are managed, then define the visibility needed for group evaluation, replica state, and waiting intervals; done requires an agreed scope and tests for the proposed behavior.