ci: revert minikube to 6GB and argo/airflow to 2 workers

4GB minikube was too aggressive — argo had 5 deployer test failures
(pod scheduling/OOM) and airflow ran for 4+ hours with 1 worker.
Keep the other improvements: fixed SHA, reruns, pod cleanup.

Author: Nissan Pow

.github/workflows/ux-tests.yml

@@ -102,10 +102,10 @@ jobs:
             workers: 4
           - backend: argo-kubernetes
             services: "minio,postgresql,metadata-service,argo-workflows"
-            workers: 1
+            workers: 2
           - backend: airflow-kubernetes
             services: "minio,postgresql,metadata-service,airflow"
-            workers: 1
+            workers: 2
           - backend: sfn-batch
             services: "minio,postgresql,metadata-service,localbatch,ddb-local,sfn-local"
             workers: 2
@@ -138,7 +138,7 @@ jobs:
         with:
           driver: docker
           cpus: 2
-          memory: 4096
+          memory: 6144
 
       - name: Restore minikube image cache
         id: image-cache