I'm not sure why the plasma arc vaporizes the material of the tap and leaves the threads in the parent metal intact, but that is what happens virtually every time in my experiance.

Maybe the remaining tapping oil shields the parent stock; as it vaporizes it forms a gas barrier? That's an interesting science question actually.

I have done this with an O/A torch. The reason it works is that the tap is a separate piece of material, so the heat transfers slower to the threaded area. If you wait too long the surrounding area will get hot enough to burn too. This method works best when you can heat it, and blow it in the least amount of time. The plasma doesn't need a preheat, but it still needs a short burn time to keep the heat transfer down to a minimum.

I have Tig welded a lot of spider cracked material. The cracks show up very distinctly as an area is heated, the material on the other side of the crack is still much colder. I have welded up stainless parts that looked like they only had a crack or two - until it was heated. Then you could see that in fact it was more like six to ten.