A data engineer must run a basic Python ETL script against a dataset of approximately 8 GB with the lowest possible compute footprint. The team wants to avoid managing Apache Spark clusters and does not need horizontal scaling. Which AWS Glue authoring approach best fits these requirements?