Summary This PR introduces a new "stepwise" assertion format that presents assertion failures as numbered sequential steps, making them more readable and parsable by LLM/AI systems. The i...