fix: retry AgentRuntime CR creation for webhook readiness in e2e

Wrap AgentRuntime CR apply in Eventually with 1m timeout and 5s poll,
matching the retry pattern already used by authbridge and combined e2e
tests. Fixes flaky webhook connection-refused race when the validation
webhook TCP listener isn't accepting connections yet despite endpoint
being registered.

Assisted-By: Claude (Anthropic AI) <noreply@anthropic.com>
Signed-off-by: Ian Miller <milleryan2003@gmail.com>

Author: r3v5

kagenti-operator/test/e2e/e2e_test.go

@@ -1182,9 +1182,11 @@ rules:
 			Expect(err).NotTo(HaveOccurred())
 			Expect(utils.WaitForDeploymentReady("runtime-agent-target", agentRuntimeTestNamespace, 2*time.Minute)).To(Succeed())
 
-			By("creating AgentRuntime CR")
-			_, err = utils.KubectlApplyStdin(runtimeAgentCRFixture(), agentRuntimeTestNamespace)
-			Expect(err).NotTo(HaveOccurred())
+			By("creating AgentRuntime CR (with retry for webhook readiness)")
+			Eventually(func() error {
+				_, err := utils.KubectlApplyStdin(runtimeAgentCRFixture(), agentRuntimeTestNamespace)
+				return err
+			}, 1*time.Minute, 5*time.Second).Should(Succeed())
 
 			By("verifying kagenti.io/type=agent on workload metadata")
 			Eventually(func(g Gomega) {