Write a Fuji-san solver

[trevorld]

There are only (at most) (28 C 4) = 20,475 possible states of monk position on a given Fujisan board (i.e. ignoring pawn suit which doesn't matter). If we build a directed network where each node is a state and we build the directed edges from each node to all the other node feasibly reached in one move then we should be able to use standard network analysis software to compute a shortest path (if it exists) from the starting state to winning state (i.e. Dijkstra's algorithm). Would need to account for games where we need 4 dice for possible starting move options (i.e. special code to handle possible moves for this special cases in the starting state).

Could use this then to generate random games of a certain difficulty (i.e. keep randomly generating games until hit a target number of optimal moves).