Resource-Aware Scheduling for Production GenAI with RAG running on Multicluster Cloud Kubernetes