Data-Driven Fairness Generalization for Deepfake Detection