vmcluster: improve components readiness check

[func25]

• Remove vmstorage liveness probe, as vminsert already handles routing and retries, while liveness probes can inadvertently introduce delays, DNS instability, and unnecessary disruptions.
• Reduce the default readiness probe interval to 5s (was 15s) and the failure threshold to 10 (was 3).
• Add a default minReadySeconds for vmstorage, to help stabilizing service during rollout.

[Haleygo]

Could you also help adjust the values for other components like vminsert, vmselect, and vmauth? 15s seems too long for all of them.

helm-charts/charts/victoria-metrics-cluster/values.yaml

Lines 107 to 114 in 499e4cd

	probe:
	# -- VMSelect readiness probe
	readiness:
	httpGet: {}
	initialDelaySeconds: 5
	periodSeconds: 15
	timeoutSeconds: 5
	failureThreshold: 3

helm-charts/charts/victoria-metrics-cluster/values.yaml

Lines 386 to 394 in 499e4cd

	# -- Readiness & Liveness probes
	probe:
	# -- VMInsert readiness probe
	readiness:
	httpGet: {}
	initialDelaySeconds: 5
	periodSeconds: 15
	timeoutSeconds: 5
	failureThreshold: 3

[func25]

@Haleygo, also updated the vmbackupmanager. LMK if you want to revert this change.

[daniel-lee-sb]

This is really needed 👍 Thanks for the quick update.

[Haleygo]

cc @AndrewChubatiuk