add side-by-side autonomous builder simulations, one using ObjectiveAI judgment

[WiggidyW]

add this to the main repo here

add a variety of various autonomous sandbox builders that build software, these will function as benchmarks

one builds using ObjectiveAI judgment, the other doesn't use ObjectiveAI judgment

will need to collect detailed metrics

this will be something you execute and it runs to completion. completion is whenever the agents have decided the software is completed

this feature is high priority, must resolve ASAP after blockers are out of the way