docs: document memory-based HPA autoscaling for admission controller (1.18)

[lucchmielowski]

Summary

Kyverno 1.18 adds HPA (HorizontalPodAutoscaler) support based on memory utilization for the admission controller (#15303). Previously the Helm chart only supported CPU-based autoscaling; now memory can be used as a scaling metric.

This is not documented in the installation or scaling documentation.

What needs to be documented

Suggested location

docs/installation/scaling.md (or equivalent HA/scaling page) — add a subsection on HPA configuration for the admission controller.

Content to cover

• The new Helm chart values that enable memory-based autoscaling (check the PR for exact value names, e.g. admissionController.autoscaling.targetMemoryUtilizationPercentage)
• How to enable autoscaling in the Helm chart (both CPU and memory, and combining them)
• Recommended thresholds and min/max replica settings for production
• Example values.yaml snippet
• Interaction with PodDisruptionBudget and high-availability configuration

References

• PR kyverno/kyverno#15303 — admission-controller autoscaling based on memory utilization