The power of Prompts: Evaluating and Mitigating Gender Bias in MT with LLMs