Add Sleep/Wake Functionality

[cstanislawski]

Currently the controller lacks a way to scale workloads down to zero during idle periods.

We should introduce "sleep" and "wake" functionality to manage the replica count of pod-spawning resources based on a schedule or one-time trigger.

This feature would include:

• New Annotations:

  • lifecycle.io/sleep-at: A specific time to scale the resource down to zero.
  • lifecycle.io/wake-at: A specific time to scale the resource back up.
  • lifecycle.io/sleep-cron: A cron schedule for scaling down.
  • lifecycle.io/wake-cron: A cron schedule for scaling up.

• State Management:

  • When sleeping a resource, the controller will add a lifecycle.io/original-replicas annotation to store the replica count.
  • When waking the resource, it will restore the replica count from this annotation. If the annotation is missing, it will default to 1.

• Behavioral Logic:

  • While a resource is sleeping (scaled to zero), the controller will only process wake or delete actions. Other scheduled actions like restarts will be paused.