Trio-ViT: Post-Training Quantization and Acceleration for Softmax-Free Efficient Vision Transformer