Shuffle Losers Function (for 2v2 hypotheses)

[LydiaFrance]

Need a function for an advanced mutation rate for losers.

Testing 2v2 battles we are looking for synergistic pairings. If you have a killer duo that has come top of the Generation 1, we don't want the breeding function to apply here because it will break up all the winning pairs.

Instead in this case we want no breeding, still keep the base level mutation, and then have a loser shuffle rate which is a inverse proportional to function of the fitness.

[philswatton]

Mutation based on fitness has been added in #20, but can't be used with any crossover operation