Revert back to using redis-ratelimiting on a VM

After the discussion in #1985 (comment 1164587025) where we looked for a way forward with the rate limiting instance in kubernetes, we decided against this for now.

The increased CPU utilization in GKE is not worth having the instance in Kubernetes while it's running fine on VMs. We will instead allocate resources in getting familiar with horizontally scalable ways of running redis instances.

This issue will track the removal of the created resources: nodepools and redis instances and the rollback to use redis instances on VMS for gstg & pre.