initial sample size returns 1

[pbstark]

Somehow the calculation of the initial sample size broke when ballot polling was introduced. Presumably, calculating the incremental sample size is also wrong. And the sample size calculations should terminate if the size reaches the appropriate population size (max_cards or contest['cards']).

[pbstark]

Resolved. issue was using np.argmax(p_history <= alpha) when the risk limit was not attained, which erroneously produced a sample size of 1 instead of N. Now test whether risk limit is attained.