Agent with naive RAG support

[JiriOndrusek]

fixes #183

This PR adds a Forage support of RAG into AI agent. The example of configuration of in memory naive RAG is:

forage.ollama.agent.model.kind=ollama
forage.ollama.agent.model.name=granite4:3b
forage.ollama.agent.base.url=http://localhost:11434

#RAG

forage.ollama.agent.embedding.model.name=nomic-embed-text
forage.ollama.agent.embedding.model.base.url=http://localhost:11434
forage.ollama.agent.embedding.model.timeout=PT30S
forage.ollama.agent.in.memory.store.file.source=company-knowledge-base.txt
forage.ollama.agent.in.memory.store.max.size=300
forage.ollama.agent.in.memory.store.overlap.size=100

forage.ollama.agent.rag.max.results=3
forage.ollama.agent.rag.min.score=0.6

(also proivided as an example: KaotoIO/forage-examples#15)

PR adds configurations of:

• embeddingModel (currently only Ollama, but should be extended in future)
• embeddingStore (vectordb) of in memory store
• default RetrievalAugmentor

if agent contains configuration of embeddingModel, embeddingStore and contains mvn:io.kaoto.forage:forage-default-retrieval-augmentor then DefaultRetrievalAugmentor is created and injected into chat model.

[orpiske]

I left one comment about the headers, but overall it LGTM. Thanks!