Record cluster worker scrape failures

[nightjard]

Summary

Add an internal histogram named
prom_client_cluster_worker_scrape_failures to record failed cluster metric
collections.

Each observation contains the number of workers that failed to return metrics:

• For timeouts, it records the number of outstanding worker responses.
• For worker-reported errors, it records the known worker failure count.
• Failed collections continue to reject without returning partial metrics.
• New observations are exposed on the next successful clusterMetrics() call.

A histogram preserves information from consecutive failures instead of only
retaining the most recent value, as a gauge would.

Testing

Added tests covering:

• Multiple workers timing out.
• Worker-reported collection errors.
• Histogram exposure after a successful recovery scrape.
• Prometheus and OpenMetrics output formats.

[jdmarshall]

I think observing 0's when there are no workers is probably unlikely to happen but is also extra information about the class of failure we may be dealing with. As written this will never catch unclassified errors such as an NPE in the try block.

For instance, regressions in the Node API are not unknown. I work on a project that doesn't work on node 24.7-24.16, for instance, due to breaking changes in child_process.fork()

[jdmarshall]

This same change should be made in worker.js as well, please, to retain feature parity.

[jdmarshall]

Are you sure you're not double-reporting this? That should be in the default registry.