Added unhealthyPodEvictionPolicy to teleport-kube-agent helm chart (#69207)

* Added unhealthyPodEvictionPolicy to teleport-kube-agent helm chart

* updated docs

* Fixed wording of docs

* fix code review findings

* Enabled unhealthyPodEvictionPolicy in 1.27, since it's on by default in beta

* replaced forceUnhealthyPodEvictionPolicy with empty value for unhealthyPodEvictionPolicy

* Test custom value in 1.26

* Added auto value to unhealthyPodEvictionPolicy

* docs update

* Removed auto value and support for unhealthyPodEvictionPolicy in k8s <1.27

* updated helm docs

* Replaced failure with warning for k8s <1.27

* fixed missing newline

* Corrected docs

* fixed doc lint issues

docs/pages/includes/helm-reference/zz_generated.teleport-kube-agent.mdx

@@ -1055,6 +1055,23 @@ a Kubernetes PodDisruptionBudget for the agent pods.
 `highAvailability.podDisruptionBudget.minAvailable` is the minimum
 available pod specified on the PodDisruptionBudget.
 
+#### `highAvailability.podDisruptionBudget.unhealthyPodEvictionPolicy`
+
+| Type | Default |
+|------|---------|
+| `string` | `"AlwaysAllow"` |
+
+`highAvailability.podDisruptionBudget.unhealthyPodEvictionPolicy` allows
+eviction of unhealthy pods even if it falls below minAvailable. When all
+pods are unhealthy (e.g. join token missing, CrashLoopBackoff, can't reach
+server due to firewall/networking), the kubernetes default policy
+(IfHealthyBudget) prevents draining a node. This chart's default
+(AlwaysAllow) allows failing pods to be killed when nodes need to drain.
+The value will be omitted with a warning on Kubernetes before 1.27 due to
+API support. Set to "" to suppress the warning at helm install time.
+See also: [unhealthyPodEvictionPolicy in Kube
+docs](https://kubernetes.io/blog/2023/01/06/unhealthy-pod-eviction-policy-for-pdbs/#how-do-i-use-it)
+
 ## `podMonitor`
 
 `podMonitor` controls the PodMonitor CR (from monitoring.coreos.com/v1)