Josh/simulated annealing

[jkarns275]

Adds simulated annealing to the island speciation strategy. In short, before a genome is considered for insertion there is a probability of a random genome being deleted from the island meaning the genome will automatically be accepted into the island regardless of its performance.

The probability is given by a function, the way I have implemented it is a function over genome number.

There are four policies included:

• default (e.g. simulated annealing is effectively disabled)
• inverse exponential = (1 + genome_number)^(-alpha)
• sin = a sinusoidal curve with a given period and range
• linear = linearly interpolates between two values for some period of time, including an initial burn in period

Note that this PR is contingent upon the other DNAS PR #41

Tested on my macbook and soon to be the cluster