feat: Implement zero-copy AsyncVectorEnv via shared memory for parallel execution

[shivansh023023]

Pull Request: feat: Implement zero-copy AsyncVectorEnv using shared memory

Description

This PR introduces AsyncVectorEnv to open_spiel/python/vector_env.py, enabling true parallel execution of environments across multiple CPU cores.

Key Architectural Details

• Zero-Copy Shared Memory: To avoid severe Inter-Process Communication (IPC) pickling bottlenecks when transmitting large observation tensors (e.g., in Chess or Backgammon), this implementation utilizes multiprocessing.sharedctypes.RawArray.
• Direct Memory Writing: Worker processes instantiate the C++ environments natively and write time_step.observations["info_state"] directly into pre-allocated shared NumPy views.
• Lightweight Pipes: The multiprocessing.Pipe is strictly reserved for small control payloads (rewards, done flags, step types), completely eliminating the need to serialize massive multi-dimensional arrays.
• Performance: In a 16-worker benchmark on chess running 1,000 steps, this shared memory architecture achieved a ~2.87x speedup over the synchronous SyncVectorEnv, fully bypassing the IPC overhead that typically throttles complex state transmission.

[lanctot]

Thanks!