You should not take into consideration anything except classical theory and what have gotten objective proof.
And should take into consideration non-types factors and a lot of mistypings due to objective low cross-typing match.
Subtypes... Any hypothesis which is far from the core theory has high risk to be wrong. By using such bs you'd rationalize a lot of wrong typings. Reinin, subtypes, functional signs, ... - their place is in a trashbin.
If you want to understand a human wider to explain the differences inside people of the same type, there are better factors to take into account - more developed, closer to known theories, more popular. For example, psychiatry explains some shifts of a behavior. Even popular here Enneagram may to have more sense than Gulenko's subtypes.
Forget the heresy! Return to the fold of the True Socionics!
Don't mess Socionics with nothing taken from nowhere.