[greendude29]

This is essentially the same solution I came up with. I'm not sure it needs to be optimized assuming we can set the starting state and the strategy for every prisoner.