Add GVL wait time metrics to Sidekiq structured logs

What does this MR do and why?

Move GVL instrumentation from ServerMetrics into InstrumentationLogger so that gvl_thread_wait_s and gvl_process_wait_s flow into the structured job completion logs via the standard instrumentation pipeline.

ServerMetrics now reads GVL values from the instrumentation hash for Prometheus, consistent with how it consumes gitaly/redis/elasticsearch metrics.

References

See gitlab-com/gl-infra/tenant-scale/tenant-services/team#384 (closed)

Screenshots or screen recordings

Before	After

How to set up and validate locally

1. In gdk.yml, make sure you have the prometheus and the sidekiq exporter enabled (run gdk reconfigure && gdk.restart after making these changes):

   # gdk.yml
   prometheus:
     enabled: true
   gitlab:
     lefthook_enabled: false
       rails_background_jobs:
         sidekiq_exporter_enabled: true

   If you are using gdk.test as your GDK host, edit config/gitlab.yml to use localhost instead (and gdk restart rails-background-jobs afterwards):

    sidekiq_exporter:
      enabled: true
   -  address: gdk.test
   +  address: localhost
      port: 3807

2. In rails console enable the feature flag

   Feature.enable(:enable_sidekiq_gvl_metrics)

3. Tail sidekiq

   tail -f log/sidekiq.log | grep gvl

   See the results streaming in.

4. Visit http://localhost:3807/metrics and observe the GVL metrics there

5. Next, disable the flag

   Feature.disable(:enable_sidekiq_gvl_metrics)

6. Observe that when tailing, the results eventually stop showing up in

   tail -f log/sidekiq.log | grep gvl

   and on http://localhost:3807/metrics

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.