A data team already maintains Apache Airflow DAGs with dependencies, retries, and scheduled backfills. It wants to run them on a managed AWS service without operating the Airflow scheduler infrastructure itself. Which service best fits?

1. An S3 bucket alone with DAG files, without any service running the scheduler.
2. A single EventBridge schedule with no workflow runtime, expecting it to execute every Airflow DAG construct.
3. Amazon Data Firehose as the execution engine for arbitrary Airflow operators.
4. Amazon Managed Workflows for Apache Airflow.