Improve Concurrency

[alisoliman]

• Added AsyncAzureOpenAI client for non-blocking LLM calls
• Converted 7 blocking endpoints to async (analyze_image, analyze_image_custom,
  enhance_image_prompt, protect_image_prompt, generate_image_filename in images.py;
  enhance_video_prompt, generate_video_filename in videos.py)
• Wrapped blocking I/O operations in asyncio.to_thread() (Cosmos DB calls, requests.get/post in
  image_pipeline.py and gpt_image.py)
• Fixed indentation errors in sora.py and videos.py async streaming code