Which limitation of a traditional data lake is directly addressed by the data lakehouse architecture?

Traditional data lakes lack ACID transactions and schema enforcement; leading to 'data swamps' with unreliable data; data lakehouses solve this by layering open table formats (Iceberg; Hudi; Delta) over the object storage files. The incorrect options ("Lack of support for storing unstructured files such as images and video", "Inability to store data from multiple cloud providers simultaneously", "Excessive storage costs caused by proprietary columnar compression formats") are distractors that don't fully capture the concept described.