It might even be a blessing if the launch mount can't be repaired in a cost-effective way and they have to start over. But an iterative, Agile-like approach looks great when perfecting welded joints. But doing too many more iterations to get to orbit, in which boosters and Starhips are expended, could break the project. Agile is appropriate for a lot of software projects (n.b. not Scrum), but maybe not for the biggest rocket ever made.
Secondly, no decision at this point will save them two months. They are in for many months of reconstruction, apart from all the other design changes needed. They are also in for FAA scrutiny that will take at least several months. There is a large incentive to be unrealistic about this: A realistic view of delays makes an already very optimistic schedule for supplying a Starship-based lunar lander to the Artemis project look very risky.
What we are seeing is the limits to scaling an iterative project. It is transitioning from learning how to weld stainless steel, which can evolve quickly, to learning how to fix multiple large problems when every iteration costs 39 rocket engines, 200 tankers of fuel/oxidizer, etc.