Reasoning About Persuasion: Can LLMs Enable Explainable Propaganda Detection?