FedFair^3: Unlocking Threefold Fairness in Federated Learning