update analysis

+analysis results

-unecessary lock file

+statistics

implement hierarchy alternative

remove generic agent

enable manager agents

enable all requests

fix complexity calculation, improve prompts and logging

improve csv output to keep plans, intermediate results, etc

store results in csv at each iteration

remove unnecessary break operation

replace the agents with restaurant ones, update the list of requests

update todo

+performance metrics

update .gitignore

use LLM to improve agent selection

refactor to simulate results with the help of llm

refactor planning

+basic multiagent implementation

+logger decorator

+gitignore

+llm call fn

update README.md