feat: Add persistent conversation memory using vector database

[LifeJiggy]

✨ Persistent Conversation Memory with Vector Database

This PR introduces long-term memory to Qwen-Agent by leveraging vector database storage, enabling context retention across sessions and significantly enhancing conversational continuity.

---

🚀 Features & Improvements

• New Feature: ConversationMemory class powered by ChromaDB for efficient vector-based storage and retrieval.
• Agent Upgrade: Added memory_cfg parameter to both Agent and Assistant classes for easy configuration.
• Persistence: Store and retrieve conversations using vector similarity, ensuring relevant context is always available.
• Testing: Comprehensive test coverage for new memory capabilities.
• Documentation: README updated, plus practical usage example.
• Dependencies: chromadb added as an optional dependency (backward compatible).

---

🧑‍💻 Usage Example

from qwen_agent.agents import Assistant

bot = Assistant(
    llm={'model': 'qwen-turbo', 'model_type': 'qwen_dashscope'},
    memory_cfg={'persist_dir': './memory_db', 'collection_name': 'conversations'}
)

# Conversations are now remembered across sessions!

---

🌟 Benefits

• Seamlessly maintains context in long or multi-session conversations.
• Improves agent continuity and overall user experience.
• Robust and production-ready, with proper error handling.
• Fully optional: no breaking changes, easy opt-in.

---

🧪 Testing

• All existing tests pass.
• New tests for persistent memory (test_conversation_memory).
• Example script included: examples/assistant_with_memory.py.

---

📦 Installation

pip install "qwen-agent[memory]"

---

Feedback, questions, and suggestions are welcome!