[Controller] Enable auto-scaling by reconciling HorizontalPodAutoscalers from Helm chart

This MR reconciles HPAs from Helm template, allowing use to account for Deployment's .spec.replicas being updated by the HPAs.

For testing:

1. Get a shell from one of the registry Pods and run the following command to increase the CPU usage: dd if=/dev/zero of=/dev/null | dd if=/dev/zero of=/dev/null | dd if=/dev/zero of=/dev/null | dd if=/dev/zero of=/dev/null &
2. Watch conditions of registry HPA. The status of different condition types must be True.
3. Monitor registry Pod events.
4. Delete the Pod that has the high CPU usage. The HPA will eventually wind down the excess Pods.

Closes #94 (closed)