Add metrics based horizontal autoscaling support for NATS cluster

[sonttran]

As we already have nats-operator to scale the number of instance of NATS cluster up and down in k8s, it would be great to be able to add metrics based horizontal autoscaling support so NATS cluster can increase/decrease number of instances horizontally on its own based on load.

Something like:

apiVersion: nats.io/v1alpha2
kind: NatsCluster
metadata:
  name: nats-cluster
spec:
  version: "2.1.8"
  size: 3 # default cluster instance
  max: 100 # max cluster instance
  min: 3 # min cluster instance
  targetCPUUtilizationPercentage: 75 # will spin up new instance if pod hitting this threshold
  pod:
    resources:
      limits:
        cpu: "200m"
        memory: "500Mi"
      requests:
        cpu: "100m"
        memory: "100Mi"

Thank you the team for the work so far on NATS. Great product!

cc @wallyqs