Add optional Prometheus metrics endpoint for runner scale set statistics

[ccang-riot]

This PR adds Prometheus metrics for runner scale set statistics and exposes them via /metrics when enabled.
Included metrics:

total available jobs
total acquired jobs
total assigned jobs
total running jobs
total registered runners
total busy runners
total idle runners

Implementation details:

metrics are opt-in via ENABLE_METRICS
metrics endpoint address is configurable via METRICS_ADDR
metrics polling interval is configurable via METRICS_POLL_INTERVAL
metrics use a dedicated Prometheus registry instead of the global registry
metrics include a runner_name label to identify the runner scale set

[ispasov]

I would suggest moving the metrics related code to a separate file (and type). This would keep the main file relatively small and the metrics ligic can be concentrated in one place.

In addition to that the start of the metrics sever together with the metrics polling can be done in the same func (for example called Start).

[ccang-riot]

Thanks for the suggestion! I’ve refactored the metrics implementation accordingly.

Moved all metrics-related logic into a dedicated metrics package (pkg/metrics/metrics.go)
Introduced a Start function that encapsulates both:

• the metrics HTTP server
• the metrics polling loop

Simplified main.go so it only handles orchestration and conditionally enables metrics via ENABLE_METRICS

This keeps main.go smaller and centralizes all metrics behavior in one place for better separation of concerns.

[ispasov]

Two things I've missed the first time:

1. Let's update the readme with these new env vars explaining what they are
2. Let's add them to the example .env file

[ccang-riot]

Updated! 🎉

[ispasov]

Tested the change and it works perfectly.
Let's update the readme and the .env example file and we are good to go.

[ispasov]

Thank you for your contribution.
This looks great!