Efficient LLM Workflows with LangChain Expression Language