Toward Accurate Vertical Federated Learning via Early Exit and Election Knowledge Distillation