using bincount instead of histogram for integration

this removes more code from the inner loop. It is more efficient,
but is it also more readable? Probably not.