Deploy ratelimiting cluster to production

Milestone goals:

• Build the production cluster that will replace redis-ratelimiting.
  • This includes prerequisite steps like choosing node size and configuration settings, based on the memory and CPU usage and configuration of the non-cluster instance.
• Migrate production redis-ratelimiting to the new cluster.
  • For ratelimiting, this does not require pre-warming or a dual-writing strategy. Just begin empty.
• Observe ratelimiting in production.
  • Check health indicators, from client and server perspectives.
  • Characterize any significant differences in resource usage between the cluster versus the old non-cluster instance.
  • Check for client latency differences, particularly in code paths that required changes to avoid running cross-slot commands.

After this milestone, we will have a "wrap up" milestone, covering an observation window and then decommissioning the old redis-ratelimiting infrastructure.