Implement Randomized Backoff for ClusterSummary Deployment Failures

[gianlucam76]

This PR introduces a dynamic, randomized backoff mechanism for ClusterSummary reconciliations. Currently, when a deployment fails, the controller retries using a static normalRequeueAfter interval.

The new logic observes the consecutiveFailures count across all features (Helm, Resources, Kustomize) and scales the requeue interval accordingly.

[vaibhavd21]

Thank you

[vaibhavd21]

This is better over MaxConsecutiveFailures

	// The maximum number of consecutive deployment failures that Sveltos will permit.
	// After this many consecutive failures, the deployment will be considered failed, and Sveltos will stop retrying.
	// This setting applies only to feature deployments, not resource removal.
	// This field is optional. If not set, Sveltos default behavior is to keep retrying.
	// +optional
	MaxConsecutiveFailures *uint `json:"maxConsecutiveFailures,omitempty"`