Beyond BLEU: GPT–5, Human Judgment, and Classroom Validation for Multidimensional Machine Translation Evaluation | Synapse