Add MCTS engine implementation and benchmark legal moves script improvements

- Introduced a new Jupyter notebook for MCTS engine implementation, including neural network architecture, MCTS node structure, and engine logic.
- Implemented MCTS engine with batch evaluation and move selection strategies.
- Enhanced benchmark_legal_moves.py to stabilize CPU frequency during measurements and allow multiple iterations for more reliable timing results.
- Updated CPU affinity settings to use a less busy core and removed unnecessary Windows-specific priority settings.

Author: miskibin

draughts/move.py

@@ -108,6 +108,10 @@ def __add__(self, other: Move) -> Move:
         move._len = len(new_captured) + 1
         return move
 
+    def __hash__(self) -> int:
+        """Hash based on the move path (start and end squares)."""
+        return hash((self.square_list[0], self.square_list[-1]))
+
     @classmethod
     def from_uci(cls, move: str, legal_moves: Iterable["Move"]) -> "Move":
         """