Added logging module & support for auto-switching to gpt-3.5-turbo-16k if the context limit of gpt-4 is reached.
