Text this: Evaluating LLMs for Code Generation in HRI: A Comparative Study of ChatGPT, Gemini, and Claude