Handle cases with more than 1 winner

[mberk]

Currently we assume that the implied probabilities sum to 1 but there are cases where they should sum to 2 or more. For example, a market betting on which horses will finish in the top three places in a race

[MikeRand]

Theoretically, wouldn't you'd need complete markets to make sure that the (now linear combinations of Arrow-Debreu) bets don't allow for arbitrage among them? For example, you could probably apply this new modified Shin to solve a podium top 3 bet if you simultaneously solved the win, 2nd, and top 3 bets.