When a user overrides values in the modeler, there has been no way to back out of it. Once the numbers were overridden, they stayed overridden.
This adds a remove modeling action that undoes those overrides and returns the plan to its underlying values.
The modeler exists so a broker can ask “what if.” They override a value, look at the result, and see whether the story holds up.
Without an undo, every one of those experiments is a one-way door. The practical consequence is that people either avoid modeling on a real quote, or they end up rebuilding to get back to clean numbers. Neither is what we want.
This closes that gap: try something, and if it isn't right, take it back off.
The Jira card is intentionally brief and the implementation is still in flight, so the specifics below are not yet confirmed and should be verified against the build before this page is used to train anyone:
Check with Dale, then replace this box with the real behavior and a screenshot.