fix: schedule initial workload and filter on worker nodes

test/e2e/plugin-config-scoring-strategy/init-workload.yaml

@@ -0,0 +1,13 @@
+apiVersion: v1
+kind: Pod
+metadata:
+  name: init-workload
+  annotations:
+    multicluster.admiralty.io/elect: ""
+spec:
+  containers:
+    - name: pause
+      image: gcr.io/google_containers/pause
+      resources:
+        requests:
+          cpu: 1

test/e2e/plugin-config-scoring-strategy/test.sh

@@ -24,8 +24,11 @@ plugin-config-scoring-strategy_test() {
   j=$2
   k $j label node --all a=b --overwrite
 
-  # Find the node with the highest CPU utilization
-  nodes=$(k $j get nodes -o name | sed s/"node\/"//)
+  k $i apply -f test/e2e/plugin-config-scoring-strategy/init-workload.yaml
+  k $i wait pod init-workload --for=condition=PodScheduled
+
+  # Find the worker node with the highest CPU utilization
+  nodes=$(k $j get nodes -l '!node-role.kubernetes.io/control-plane' -o name | sed s/"node\/"//)
   most_allocated=""
   most_cpu_requests="0"
   for node in $nodes; do