Support scraping metrics from multiple instances

When running multiple spectroperf instances on the same machine the second will fail with:

```
2025/08/05 09:32:26 listen tcp :2112: bind: address already in use
```

This is the port that we are trying to use to expose the prometheus metrics. We should make the selection of the port being used resilient to this error and update the prometheus config file to scrape all the possible endpoints

Locate the code that starts the Prometheus metrics listener on :2112 and the Prometheus configuration that defines scrape endpoints. Reproduce the conflict with two spectroperf instances, then verify that a second port is selected and all instance endpoints appear in the generated configuration.

Written by the indexing model from the issue text.