fix: cleanup orphaned components when parent cluster is missing

[wallyxjh]

Problem

PR #18 fixed orphaned Component cleanup for the Terminating case (Components with deletionTimestamp), but did not cover the Running orphan case.

When a Cluster is deleted but its owned Components still exist without a deletionTimestamp, the componentLoadResourcesTransformer enters an infinite 1-second requeue loop:

build error: requeue after: 1s as: failed to get cluster test-db: 
clusters.apps.kubeblocks.io "test-db" not found

This happens when Kubernetes GC fails to complete cascade deletion (e.g. due to timing, force delete, etc.). These orphaned Components waste controller resources and cause log spam across all affected namespaces.

Root Cause

In transformer_component_load_resources.go, when the Cluster is not found, the code unconditionally returns newRequeueError(requeueDuration, err.Error()) without checking whether the Component is an orphan:

cluster := &appsv1alpha1.Cluster{}
err = transCtx.Client.Get(..., cluster)
if err != nil {
    return newRequeueError(requeueDuration, err.Error()) // infinite 1s retry
}

The componentDeletionTransformer (fixed in PR #18) is skipped entirely because these Components have no deletionTimestamp.

Solution

Added orphan detection in componentLoadResourcesTransformer:

1. When Cluster returns NotFound, check if the Component has an ownerReference pointing to that Cluster (confirming it is an orphan).
2. If orphaned, trigger deletion via client.Delete() to set deletionTimestamp.
3. On the next reconcile, componentDeletionTransformer (with PR fix: handle component deletion when parent cluster is missing #18 fix) handles the actual cleanup.

Changes

• controllers/apps/transformer_component_load_resources.go: Add isOrphanedComponent() and handleOrphanedComponent(), inject client.Client for direct API calls.
• controllers/apps/component_controller.go: Pass Client to componentLoadResourcesTransformer.