Feat/advisor tools (#38)

Feat/vllm endpoints (#36)

Fix/structured json output (#30)

reconfiguring llm classes to match the persona changes for response generation

update llm client for better responses

Changes to include single model represented as multiple personas