Text this: Enhancing Deepfake Detection Through Quantum Transfer Learning and Class-Attention Vision Transformer Architecture