Text this: Competency of Large Language Models in Evaluating Appropriate Responses to Suicidal Ideation: Comparative Study