v0.5.1

Simplifying adaptive_simpsons_rule so that with each iteration N doubles. This causes accuracy to more closely match the tolerance argument given and it seems to make it more efficient (less computation time for a given level of accuracy).