Revisiting Lower Bounds for Two-Step Consensus