PTQ4ViT: Post-training Quantization for Vision Transformers with Twin Uniform Quantization