I think I see the error here. (It might be my communication.) The skills needed to prevent flaws are different than the ones used on the fly when finding flaws. At least, that is my take.
The skills need to prevent flaws are predicated on the ones to find flaws. If I'm preventing flaws, its because I can already recognize them. You cannot prevent flaws (effectively) if you cannot find them.
Maybe you are misunderstanding. I am not dissing design work at all. I am saying, that when a GM uses an AP, they use and sometimes practice different skills that they would not use in adventures of their own design. Hence, my original claim, that APs can teach young and experienced GMs alike. It can create new tools and/or remind a GM of tools they haven't used in a long time.
There are two things going on here. Firstly, the ability to recognize flaws that need to be fixed isn't something that is special to APs. You've introduced that APs are special because they have more(?) chances of being flawed and therefore a GM learns more about recognizing flaws. This isn't clear because work done by a GM for their own games has already done this pass -- there are fewer flaws to recognize in play because the GM has already done most of this work in design. There's nothing unique about APs, except maybe the higher likelihood of poor design for a given GM's game.
Secondly, you're advocating for APs as teaching new tools. This is flawed because you must already be good enough at GMing to recognize when an AP is doing something clever and worth learning and when it's doing something badly than needs correction. The skill to recognize these things is not taught by APs, it's actually harmed by them, because there's the assumption that the AP is actually well-designed and so emulating all of it is something you should do. And, perhaps it is, for a given table, but there's nothing in the AP that teaches this especially over things generated by a GM. What I mean here is that it's just as likely for a GM to learn that things do or don't work with their own material as it is with an AP.
The presumed value of an AP is that the GM has less work to do to run a game that is presumably well-designed. I find most APs are full of very specific approaches that don't suit a number of tables, and that this isn't apparent at all. There's a reason why there are so many threads on fixing APs and/or blogs that do the same -- they dissect the adventure with the eye of an experienced GM and show how the adventure is poorly designed in places and how it might be modified. If the APs taught this easily, as you claim, then the need for these threads/blogs would be reduced and they wouldn't be as popular as they are. Except, they proliferate.
This is exactly my point. The difference between your creation and an AP is different, and sometimes those differences use different skill sets. So the experienced GM that has used their own material for twenty years might experience something brand new (or relived or re-experienced) when running an AP as is.
Yes, a poorly designed game that shows the GM why they long ago learned to not do that. Like the opening of Descent, where the PCs are pressganged during a scene where an NPC gets a whole scene as to how awesome they are, and then threatened with death unless they do the adventure for the NPC. This is absolutely terrible design -- the worst of railroading and Force. It utterly strips any attachment you might have built with the players to the adventure and makes it adversarial. And, the fun thing is that the adventure is written as if the players as supposed to like this, and be friendly with the NPC threatening them and press-ganging them. It's absolutely terrible! I learned nothing from this, because I had to completely rewrite the start of this AP so that it doesn't immediately send my players into adversarial mode.
APs are not, in any way, unique or special in how they can let a GM learn. Most GMs learn more when they go off and do their own stuff, and not from APs. Is there a possibility a GM can learn something from an AP? Sure. But, they aren't particularly good teachers, on average.