Service Health Probe Improvements

### GA ETA: Q2 2026*
*ETAs are estimations and are subject to change

Previously, the behavior of SLB probes in Kubernetes was complex, relying on node connectivity, kube-proxy connectivity, and application state, which made troubleshooting challenging. This complexity was especially problematic during updates, such as the 1.24 upgrade, where changes in the cloud provider impacted the functionality of ingress-nginx, causing pods to appear ready without receiving traffic due to misconfigured SLB probes. The old system required correct configurations on both pod readinessProbes and service annotations, leading to customer confusion and redundant efforts.

The new behavior simplifies the SLB probes to depend solely on node connectivity and kube-proxy connectivity, rendering the application health probes by Kubernetes sufficient to determine pod readiness. This change implies that the SLB probes are almost irrelevant for most users, and there is a consideration to potentially deprecate the service annotations entirely unless a specific need arises.