Private Equilibrium Release, Large Games, and No-Regret Learning