The article is an interesting tutorial on solder mask indeed. However, it does not discuss why we apply add a clearance to the solder mask at all.
The purpose of the solder mask is to protect the board and assist in the assembly process. The ideal solder mask is one without clearance, where the solder mask opening matches the pad exactly. This is what the designer ideally would want. The problem is that the fabricator cannot deliver this: there are unavoidable deviations in the fabrication process. The cost of the deviation is very assymetric: mask on the pad is a disaster, a clearance around the pad is no big deal. Hence the need for a clearance. Another issue is the minimal width. Appyling a larger clearance aggravates the minimum width problem. The ring to be as small as possible, but no smaller.
The question is how much clearance to apply. This is very complicated: it depend on the equipment, the process, the mask type, the copper weight, even the mask color. The capabilities published by a fabricator are a crude simplification of a complex reality. The designer may not even know where the board will be fabricated. In short, the designer cannot know what the ideal clearance is. Whatever clearance he applies, the only thing he knows for sure is that it is not be the optimal one. The only person that knows the ideal clearance is the fabricator, at the moment of fabrication.
The fabricators have a CAM system and are quite capable to optimize a mask. Complicated stuff must be done to get to the required via protection. As quite a few masks are wrong, they are used to remove the applied compensations, and re-apply them to their requirement. This is one of the things CAM engineers do to make a living.
Hence, the best clearance to apply is no clearance at all. Let the fabricator apply the clearance that best suits his process. If you want to have minimal problems, send the fabricator the mask without extra clearance, and let him compensate for the deviations in his process. After all, he applies other changes to compensate for deviations in fabrication: copper is spread to compensate for over-etching, and layers are scaled to compensate for distortions in the lamination press. We do no not expect the designer to compensate for these, do we? If we don’t to it for etching and lamination, why do it for the solder mask?