Implement calculate_error and generate_population methods. Issue #13

Fix comparison between particle fitness and pbest/gbest. Issue #13

Add initial PSO implementation.

Move preprocessing module into tools package.

Move backpropagation algorithm into trainers package. Issue #11

Implement training framework. Issue #10

Allow caller to pass bias as a numeric value instead of a boolean. Issue #14

Move bias check into _concat_bias.

Remove extraneous parameter from _concat_bias. Issue #9

Fix false bias bug. Issue #9

Add integration test for false bias bug.

Add test for arbitrary mlp layers.

Add ability to construct MLP networks with arbitrarily-many layers.

Allow MultilayerPerceptron to accept network shape parameter instead of separate layer node counts.

Generalize code in __recall to work on arbitrarily many layers.