feat(daemonset): add restart and rollout-aware wait methods

[rnetser]

Problem

DaemonSet is missing two capabilities that force consumers to shell out to oc:

1. No restart method — oc rollout restart works by patching the pod template with a kubectl.kubernetes.io/restartedAt annotation, triggering a rolling update. The wrapper has Resource.update() which can do a merge-patch, but there's no convenience method on DaemonSet.

2. No rollout-aware wait — wait_until_deployed only checks desiredNumberScheduled == numberReady, which is racy after a restart (old pods are still ready, returns immediately).

Root Cause

File: ocp_resources/daemonset.py

• No restart() method exists. VirtualMachine has one (via subresource API), but DaemonSet restart is a standard annotation patch — not currently implemented.
• wait_until_deployed (line 15) checks only:

  if desired_number_scheduled > 0 and desired_number_scheduled == number_ready:

  This doesn't verify pods are actually updated — just that enough are ready.

Proposed Approach

restart() method

Patch the pod template with a restartedAt annotation (same as oc rollout restart):

def restart(self):
    self.update(resource_dict={
        "spec": {"template": {"metadata": {"annotations": {
            "kubectl.kubernetes.io/restartedAt": datetime.utcnow().isoformat()
        }}}}
    })

wait_for_rollout() method

Check all three conditions (mirrors oc rollout status):

1. status.observedGeneration == metadata.generation
2. status.updatedNumberScheduled == status.desiredNumberScheduled
3. status.numberAvailable == status.desiredNumberScheduled

Similar to how Deployment.wait_for_replicas checks updatedReplicas.

Done

• Add restart() method to DaemonSet class
• Add wait_for_rollout() method to DaemonSet class
• Keep existing wait_until_deployed unchanged (no breaking change)
• Add tests for both new methods