Validates that engagement scoring, explainability weights, and export tooling are behaving as expected against the loaded mock dataset.