Trade-off between Payoff and Model Rewards in Shapley-Fair Collaborative Machine Learning