A Unifying Perspective on Multi-Calibration: Game Dynamics for Multi-Objective Learning