One Prompt To Rule Them All: LLMs for Opinion Summary Evaluation