Text this: Beef Carcass Grading with EfficientViT: A Lightweight Vision Transformer Approach