Advancing Automated Ethical Profiling in SE: a Zero-Shot Evaluation of LLM Reasoning