Allow customization of pod probe parameters (timeoutSeconds, periodSeconds, failureThreshold)

[cdaguerre]

Problem

The chart hardcodes livenessProbe, readinessProbe (replicas only), and startupProbe blocks in templates/statefulset.yaml and templates/deploy_valkey.yaml with only the exec command set:

startupProbe:
  exec:
    command: [ "sh", "-c", "valkey-cli ping" ]
livenessProbe:
  exec:
    command: [ "sh", "-c", "valkey-cli ping" ]

There is no way for chart users to override timeoutSeconds, periodSeconds, failureThreshold, initialDelaySeconds, or successThreshold.

Why this matters

GKE 1.35+ enforces timeoutSeconds on exec probes (previously the field was silently ignored for exec probes; the kubelet now strictly enforces the timeout). Probes without an explicit timeoutSeconds default to 1 second, which is often insufficient for valkey-cli ping under load or during startup. This trips the GKE Recommender exec-probe-timeout deprecation alert and pauses cluster auto-upgrade until the cluster has been clean for 7 days.

Reference: https://docs.cloud.google.com/kubernetes-engine/docs/deprecations/exec-probe-timeouts

The same enforcement is being adopted by other Kubernetes distributions tracking upstream KEP-1972 and KEP-2727.