A 2-step Framework for Automated Literary Translation Evaluation: Its Promises and Pitfalls