Such systems are fully flawed and meaningless. This test even claim that Socionics is "obscure" while their system is "objective". I can't agree with such claims. As far as I'm concerned Socionics (Model A) is the best among all the so-called Jungian typologies.
If you have 7000000000 types you get everyone typed correctly. The addition of subtype is just a means to cover up one's inability to accurately type correctly in the framework of the 16 types.
Each personality typology system cuts through just one facet of the complex human psyche. It all has its own scope.
Jung's typology, for example, is a dissection of human types based on I/E attitudes and the 4 psycho-cognitive functions. Obviously, there are many factors that Jung's typology cannot address. For example, IQ, moral standing, etc. are part of human traits, but these are beyond the scope of Jungian typology. Therefore, I think it is crucial to be able to accurately determine the 16 types. For other categories, we have other classifications, such as E9, Dark Traid, attachment types, etc. Each of these methods has its own role to play in describing humans from different sides. To develop subtypes too much is just to cover up one's incompetence.