Expand Integration with Langchain4j for ChatMemory and RAG

[clun]

Thank you @glaforge for the Langchain4j LLM Adapter. As of today it allows to

• use all the ChatModel from Langchain4j
• Propagate Tools configurations to the Langchain4j context for execution.

I hope this integration could be moved forward with

1. Couple ChatMemory and SessionServices to share the conversation history
2. Use ArtifactService for the Langchain4j retrievers (RAG)
3. Use MemoryServices for the Langchain4j retrievers (long context)

Context

We are running a system where users will set up agents in our user interface. From there it will create an conversational Agents with Langchain4j

In the user interface:

You can think about something like

To move to a multi-agents system we would like to be able to load those Agent Definition as LLMAgents from ADK and reuse what we have already in place