How to Split the Pie: Optimal Rewards in Dynamic Multi-Battle Competitions