NewNewVersion: More Fixes & Improvements

[chowington]

Actively-maintained repo: https://github.com/crcresearch/agentic_collab

This is an updated version of the codebase based on @drudilorenzo's great fork featuring the following improvements, on top of his own additions:

• Added a true headless mode that doesn't require headless Chrome, and a script to print the headless simulation state in real time
• Implemented OpenAI's Structured Outputs for most prompts to increase consistency
• Refactored run_gpt_prompt.py and improved prompt template logic
• Improved error logging
• Added the ability to customize the simulation world map by renaming locations and objects
• Allowed passing an agent history file into the automatic execution command
• Separated prompt templates that are actually used from those that aren't
• Removed superfluous July1_the_ville_isabella_maria_klaus-step-3-* sim folders that just took up space
• Updated .gitignore to ignore environment and movement folders within simulation folders, as they severely muddle the repo (to commit full sim runs, it's best to first compress them as described in the original repo's README)
• Added various other output processing scripts, including some NLP scripts
• Removed unnecessary duplicated runs of the hourly schedule LLM calls
• Removed hardcoded sleeping from start of day
• Improved some prompt wordings
• Replaced star imports (import *) with enumerated imports
• Fixed various other small bugs
• Other small code QOL changes

Note that due to the size of these changes (mainly deleting the huge old simulation folders), GitHub cannot display the diff correctly.