refactor(script): remove sync endpoint, add thread-pool & timeout

- Remove unused POST /script/generate sync endpoint and frontend generate()
- Move JSON parsing/validation to asyncio.to_thread() to avoid event-loop blocking
- Add 60s asyncio.timeout() around entire script generation pipeline
- Migrate volcengine_provider to unified AsyncArk client

python-api/app/api/v1/script.py

@@ -49,29 +49,6 @@ async def get_categories():
     )
 
 
-@router.post("/generate", response_model=ApiResponse[list[ScriptShot]])
-async def generate_script(request: GenerateScriptRequest):
-    """
-    同步生成脚本
-
-    直接返回生成的分镜列表，适合快速预览。
-    """
-    service = get_script_service()
-
-    shots = await service.generate_script(
-        category=request.category,
-        subcategory=request.subcategory,
-        duration=request.duration,
-        script_type=request.script_type,
-        model=request.model,
-    )
-
-    return success_response(
-        data=shots,
-        message=f"成功生成 {len(shots)} 个分镜",
-    )
-
-
 @router.post("/generate/stream")
 async def generate_script_stream(request: Request, data: GenerateScriptRequest):
     """