fix: prevent flaky InferenceGraph auth deletion tests (#1221)

Set the deployment status to Available in BeforeEach for the auth test
context. Without this, the controller enters a requeue loop with
exponential backoff (deployment never becomes available in envtest),
and the workqueue rate limiter delays can exceed the 10s test timeout,
causing intermittent failures in the SA deletion and CRB cleanup tests.

Signed-off-by: Pierangelo Di Pilato <pierdipi@redhat.com>

Author: pierDipi

pkg/controller/v1alpha1/inferencegraph/controller_test.go

@@ -1457,6 +1457,20 @@ var _ = Describe("Inference Graph controller test", func() {
 				},
 			}
 			Expect(k8sClient.Create(ctx, inferenceGraph)).Should(Succeed())
+
+			// Wait for the deployment to be created, then set its status to Available.
+			// Without this, the controller enters a requeue loop with exponential backoff
+			// (deployment never becomes available in envtest), and the rate limiter delays
+			// can exceed the test timeout, causing flaky failures in deletion tests.
+			deployment := &appsv1.Deployment{}
+			deployKey := types.NamespacedName{Namespace: inferenceGraph.GetNamespace(), Name: inferenceGraph.GetName()}
+			Eventually(func() error {
+				return k8sClient.Get(ctx, deployKey, deployment)
+			}, timeout, interval).Should(Succeed())
+			deployment.Status.Conditions = []appsv1.DeploymentCondition{
+				{Type: appsv1.DeploymentAvailable},
+			}
+			Expect(k8sClient.Status().Update(ctx, deployment)).Should(Succeed())
 		})
 		AfterEach(func() {
 			_ = k8sClient.Delete(ctx, inferenceGraph)