docs: update docs/customize/custom-llm-server.md with Moonshot AI example

- Refactor JSON structure to use array format for "messages".
- Add an official Moonshot AI example to the customization guide.
- Include a Python code snippet demonstrating how to interact with the Moonshot AI API.

This change improves the readability of the JSON configuration and provides a practical example for users to understand how to customize their LLM server with Moonshot AI integration. The Python code snippet serves as a clear guide for users to follow when setting up their API calls.

Author: phodal

docs/customize/custom-llm-server.md

@@ -114,14 +114,41 @@ And the request body will be:
 
 ```json
 {
-	"user": "12345",
-    "model": "gpt-4",
-    "messages": [{"role": "user", "message": "..."}]
-  }
+  "user": "12345",
+  "model": "gpt-4",
+  "messages": [
+    {
+      "role": "user",
+      "message": "..."
+    }
+  ]
+}
 ```
 
 ### Moonshot AI example
 
+Official Moonshot AI doc: https://platform.moonshot.cn/docs/api-reference
+
+```
+from openai import OpenAI
+
+client = OpenAI(
+    api_key="MOONSHOT_API_KEY",
+    base_url="https://api.moonshot.cn/v1",
+)
+
+completion = client.chat.completions.create(
+  model="moonshot-v1-8k",
+  messages=[ 
+    {"role": "system", "content": "你是 Kimi，由 Moonshot AI 提供的人工智能助手，你更擅长中文和英文的对话。你会为用户提供安全，有帮助，准确的回答。同时，你会拒绝一些涉及恐怖主义，种族歧视，黄色暴力等问题的回答。Moonshot AI 为专有名词，不可翻译成其他语言。"},
+    {"role": "user", "content": "你好，我叫李雷，1+1等于多少？"}
+  ],
+  temperature=0.3,
+)
+
+print(completion.choices[0].message)
+```
+
 Engine Server:
 
 ```