RFC: game_theory: Use numba.typed.Dict in _vertex_enumeration_gen

[oyamad]

I let ChatGPT read the existing code of vertex_enumeration.py and got the following suggestion:

• The current _vertex_enumeration_gen has complexity O(V_0 * V_1), where V_i is the number of vertices of the best response polytope of player i.
• One should use Dict in place of the inner loop, which will reduce the complexity to O(V_0 + V_1).

Benchmarks with random n x n games showed significant speedups, and larger speedups for larger games: about 1.3x faster for n=12, 2x faster for n=14, 15, and 4x faster for n=17. (And no additional overheads for small size games.)

[mmcky]

thanks @oyamad

[mmcky]

thanks @oyamad I have now release v0.8.1