The regression lane is being cleaned up so it can serve as a real canary system rather than a pile of historical scripts.
Current working rules:
- each regression script should set up its own environment
- path-only and path-plus-open bootstraps stay explicit
- useful historical scripts are repaired, not blindly retained
- obsolete or misleading scripts are retired
This work also supports future student and operator refresh paths, because a controlled regression lane is the easiest way to prove that documentation and runtime behavior still agree.