Learning Equilibrium with Estimated Payoffs in Population Games