Subtree crossover distribution test displays incorrect statistics

[bburlacu]

In the SubtreeCrossoverDistributionsTest, we have the following code which tests the effects of crossover:

for (int gCount = 0; gCount < generations; gCount++) {
  for (int i = 0; i < POPULATION_SIZE; i++) {
    var par0 = (ISymbolicExpressionTree)trees.SampleRandom(random).Clone(); // parents are cloned
    var par1 = (ISymbolicExpressionTree)trees.SampleRandom(random).Clone(); // parents are cloned
    SubtreeCrossover.Cross(random, par0, par1, 0.9, 100, 10); // population is never updated
  }
}

Since the parents are cloned and result of SubtreeCrossover.Cross is discarded and trees is never actually updated, the statistics displayed at the end do not reflect the effects of crossover, but just show the initial distribution returned by the tree creator.

[bburlacu]

r16292: Update the population with crossover children.

[mkommend]

r16429: Merged r16292 into stable.