Prometheus metrics are missing for sidekiq processes

Prometheus metrics don't seem to be getting scraped consistently for all sidekiq processes in a sidekiq cluster.

For example, this change is no longer working as expected: #32258 (closed)

Charting this metric over 1 day:

sidekiq_concurrency{fqdn="sidekiq-pipeline-01-sv-gprd.c.gitlab-production.internal"}	

https://prometheus-app.gprd.gitlab.net/graph?g0.range_input=1d&g0.expr=sidekiq_concurrency%7Bfqdn%3D%22sidekiq-pipeline-01-sv-gprd.c.gitlab-production.internal%22%7D%09&g0.tab=0

The problem appears to be that the results for a single pid are being returned, not all pids in the sidekiq cluster on the host.

cc @mkaeppler