fix(chatgpt): Resolve async response issue with Azure GPT-4o API

[JxQg]

Description

Fixed an issue where the Azure GPT-4o API's asynchronous responses contained empty choices arrays, leading to incorrect handling of empty data.

How Has This Been Tested?

This is my configuration.

LOCAL_DB_HOST=mysql
LOCAL_DB_PASSWORD=******
ALLOWLISTED_PLUGINS=db_dashboard
LLM_MODEL=proxyllm
PROXY_API_KEY=*
PROXY_API_BASE=https://*.openai.azure.com/
PROXY_API_TYPE=azure
PROXY_SERVER_URL=https://*.openai.azure.com/openai/deployments/gpt-4o/chat/completions?api-version=2024-02-15-preview
PROXY_API_VERSION=2024-02-15-preview
PROXYLLM_BACKEND=gpt-4o

When I successfully deploy with docker or a local conda environment, no matter what questions are raised on the front end, the output will be disconnected at the end and an exception will be thrown.As follows:

Logs:

 messages:
[{'role': 'system', 'content': '你是一个有用的 AI 助手。'}, {'role': 'user', 'content': '测试正常AzureOpenAPI响应'}]
llm_adapter: <OpenAIProxyLLMModelAdapter model_name=proxyllm model_path=chatgpt_proxyllm>

model prompt: 

system: 你是一个有用的 AI 助手。
human: 测试正常AzureOpenAPI响应

async stream output:

INFO:     172.19.0.1:58758 - "POST /api/v1/chat/mode/params/list?chat_mode=chat_normal HTTP/1.1" 200 OK
INFO:     172.19.0.1:58774 - "GET /api/v1/feedback/select HTTP/1.1" 200 OK
2024-08-02 09:34:57 a71ebe4fcaf1 dbgpt.model.cluster.worker.default_worker[1] INFO is_first_generate, usage: None
要测试 Azure OpenAPI 的正常响应，你可以按照以下步骤进行：

1. **创建 Azure 资源**：确保你有一个 Azure 订阅，并创建一个 API Management 服务或 Azure Function App 来托管你的 API。

2. **创建 OpenAPI 定义**：编写或生成你的 OpenAPI (Swagger) 定义文档。你可以使用 Swagger Editor 或其他工具来创建这个文档。

3. **部署 API**：将你的 API 部署到 Azure 上。你可以使用 Azure 门户、Azure CLI 或 Azure DevOps 等工具来进行部署。

4. **获取 API 的 URL**：部署完成后，你会得到 API 的 URL，这个 URL 将用于测试。

5. **测试 API**：使用工具如 Postman、cURL 或者编写一个简单的脚本来发送请求到你的 API，并检查响应。

下面是一个使用 cURL 测试 API 的简单示例：


curl -X GET "https://your-api-url/endpoint" -H "accept: application/json"


如果你想使用 Postman，请按照以下步骤：

1. 打开 Postman。
2. 创建一个新的请求。
3. 设置请求方法（如 GET、POST 等）和 URL。
4. 添加必要的头信息（如 `accept: application/json`）。
5. 点击“Send”按钮发送请求。
6. 检查响应内容。

### 示例 OpenAPI 定义
这是一个简单的 OpenAPI 定义示例：


openapi: 3.0.0
info:
  title: Sample API
  description: A sample API to illustrate OpenAPI concepts
  version: 1.0.0
paths:
  /hello:
    get:
      summary: Returns a greeting message
      responses:
        '200':
          description: A successful response
2024-08-02 09:35:09 a71ebe4fcaf1 dbgpt.model.cluster.worker.default_worker[1] ERROR Model inference error, detail: Traceback (most recent call last):
  File "/app/dbgpt/model/cluster/worker/default_worker.py", line 248, in async_generate_stream
    async for output in generate_stream_func(
  File "/app/dbgpt/model/proxy/llms/chatgpt.py", line 43, in chatgpt_generate_stream
    async for r in client.generate_stream(request):
  File "/app/dbgpt/model/proxy/llms/chatgpt.py", line 230, in generate_stream
    async for r in self.generate_stream_v1(messages, payload):
  File "/app/dbgpt/model/proxy/llms/chatgpt.py", line 265, in generate_stream_v1
    if r.choices[0].delta.content is not None:
AttributeError: 'NoneType' object has no attribute 'content'

Snapshots:

Include snapshots for easier review.

Checklist:

• My code follows the style guidelines of this project
• I have already rebased the commits and make the commit message conform to the project standard.
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• Any dependent changes have been merged and published in downstream modules

[csunny]

@JxQg Well, it's a very useful PR. But could you please change the comments to en?

[JxQg]

@csunny I'm so sry for comment format that does not conform to the code specification. I have modified it and push it .

[csunny]

r+

[fangyinc]

LGTM.