GAIN Game-Theoretic Design of Fair Incentive Mechanisms for Multiple Model Owners in Federated Learning