Reuse OpenAI clients to improve throughput

[luojiyin1987]

Problem

Currently, every call to ChatGPT_API, ChatGPT_API_with_finish_reason, and ChatGPT_API_async creates a new OpenAI/AsyncOpenAI client instance. This has significant overhead:

• HTTP connection pool is created on each call
• Reduces throughput in high-concurrency scenarios
• The async version uses async with which creates and closes a client on every request

Solution

• Add module-level singleton clients cached per API key
• Replace client creation with _get_sync_client/_get_async_client helper functions
• Fix chat_history mutation side effect (use list concatenation instead of append)
• Fix error return type inconsistency in ChatGPT_API_with_finish_reason
• Add missing re import

Impact

• Improved throughput for batch API calls
• Reduced connection overhead
• Better resource utilization