Detecting Conversational Mental Manipulation with Intent-Aware Prompting