bugfixing; fixed toolbox; isolated [Base|AI|Human]Message crap logic to the agent interface; implemented tests

minor refactoring

embracing langgraph; addapted LLMFactory, Args.LLMInterface, IAgent, Toolbox and agents.py; simplified graph: removed math and encryption agents while promoting the reasoner

built management agents with llama_index, recursive self call for task breakdown, fixed LLMFactory to create LLMs from different interfaces

setup project structure: getting ready to build my agents