Increase liveness/readiness probe failures for gprd
Corrective action for gitlab-com/gl-infra/production#5466 (closed)
In order to better support a pod that is trying to recover from a large wal replay, we want to double the amount of failures that can be tolerated on the liveness and rediness probe (currently 5m and 10m total respectively) in order to give the pod the greatest chance of not being marked as failed by Kubernetes and restarted.