Solving quadratic multi-leader-follower games by smoothing the follower's best response. (4th March 2022)