[meta] Shared Runners transition to GCP plan

Steps that are left:

• enable monitoring
  • gather pre-aggregated metrics on prometheus.gitlab.com (https://dev.gitlab.org/cookbooks/chef-repo/merge_requests/1463, current blocker)
  • put graphs at https://performance.gitlab.net/dashboard/db/ci
  • separate Consul+Prometheus network traffic form job related traffic
  • prepare alerting based on pre-aggregated metrics => not needed directly for the GCP migration
• prepare monitoring of GCP resources (#2585 (closed))
  • number of used instances (equivalent of what we have already for DO) => https://gitlab.com/gitlab-org/ci-cd/gcp-exporter
  • quota usage
  • costs usage and prediction => unfortunately, from what I found in API documentation, we're unable to get such data through API :(
  • tag created instances with one of: srm, gsrm, gsrm-dev, stg-srm (so we can group it later on graphs)
• make GCP runners the default ones
  • check and probably request an increase of some quotas
    • check current quota usage and estimate needs for migration to GCP
    • request quota limits increase (if needed)
    • confirm that quota limits were increased
    • prepare alerting for GCP quotas
  • increase concurrent and limit values in GCP Runners configuration
  • decrease concurrent and limit values in DO Runners configuration
  • create private-runners-manager-{3,4}.gitlab.com placed in GCP: #3696 (closed)
  • update https://about.gitlab.com/gitlab-com/settings/#shared-runners => www-gitlab-com!10276 gitlab-org/gitlab-ce!18209
• disable DO Runners and leave them as a backup: #4172 (closed)
• Change PRM runner managers to group runners on GitLab.com: #4461 (closed)
• Remove Consul cluster for CI Monitoring from GCP part of the CI infrastructure: #4407 (closed)
• Replace own cache servers with GCS for GCP Runners fleet: #4565 (closed)

/cc @ayufan @edjdev @stanhu