3 of the major bugs fixed

edited logic of follow_up

corrected reward function , grader.py , inference.py , corrected the structures acoording to guidelines , in inference.py the openai cilent is not yet loaded so it won't work for now

added an rule based py script imitating llm model , reward function and other logics are broken and needs shreya wala glue , apart from it the server is running just fine and finally the base it ready we can start writing the main mathematical logic , do not touch dockerfile