feat(library): AI coding-agent benchmark: a reproducible test of debugging, test generation, and refactoring (#8064)

* feat(library): AI coding-agent benchmark: a reproducible test of debugging, test generation, and refactoring

* Pi Babysit: address PR #8064 feedback

---------

Co-authored-by: Sim Pi Agent <pi@sim.ai>

Author: icecrasher321