Use non-managed downstream cluster in multi-cluster label tests

This makes the multi-cluster end-to-end test suite compatible with
setups in which only one, non-managed downstream cluster exists, such as
the test-Fleet-in-Rancher CI workflow.

.github/workflows/e2e-multicluster-ci.yml

@@ -185,6 +185,9 @@ jobs:
           FLEET_E2E_NS: fleet-local
           FLEET_E2E_NS_DOWNSTREAM: fleet-default
         run: |
+          # Force use of non-managed downstream cluster for portability
+          export CI_REGISTERED_CLUSTER=$(kubectl get clusters.fleet.cattle.io -n $FLEET_E2E_NS_DOWNSTREAM -o jsonpath='{range .items[*]}{.metadata.name}{"\n"}' | grep -v second)
+
           kubectl config use-context k3d-upstream
           ginkgo --github-output e2e/multi-cluster
       -

e2e/multi-cluster/not_matching_targets_delete_bd_test.go

@@ -31,7 +31,7 @@ var _ = Describe("Target clusters by label", func() {
 
 	BeforeEach(func() {
 		k = env.Kubectl.Context(env.Upstream)
-		kd = env.Kubectl.Context(env.ManagedDownstream)
+		kd = env.Kubectl.Context(env.Downstream)
 	})
 
 	JustBeforeEach(func() {