Summary

Add llama.cpp local LLM inference configuration example to config.example.yaml

What

Added commented configuration example for llama.cpp local model server:

• Uses OpenAI-compatible API via langchain_openai:ChatOpenAI
• Default port: 8080 (llama.cpp standard)
• Example uses Qwen3.6 35B model as referenced in the issue

Why

Issue #2931 requests configuration example for llama.cpp local deployment. Users deploying local models via llama.cpp can now easily configure DeerFlow without manually writing configuration.

How

Configuration follows existing OpenAI-compatible pattern:

• base_url: http://localhost:8080/v1
• api_key is optional (commented with explanation)
• Includes Docker deployment note (host.docker.internal)
• Documents server startup command and advanced options

Test plan

• Configuration follows existing patterns in config.example.yaml
• All required fields documented (name, display_name, use, model, base_url)
• api_key handling properly explained
• Docker deployment note included

Checklist

• Code follows project style guidelines
• Documentation updated (config.example.yaml)

Fixes #2931