If you do a google search for keywords like "looks" "appearance" "correlation" "link to" "personality" "studies" etc. you'll find numerous links to articles published by people working in western universities and accredited research institutions on topic of how appearance is linked to various faucets of personality.
I'm not sure why Socionics always gets the negative rep for the mentions of external traits of types, since there are so many publications in English, some coming from tor tier universities in United States, that basically cover a variety of connections between looks and personal qualities that have been shown by various studies. While I haven't seen any studies that address socionics types in particular (or maybe no translated studies), it wouldn't be an unreasonable idea that socionics type also leaves some visual imprints that can be discerned after some careful observation.
How much attention gets drawn to this also depends on which Socionics author you read. Some authors like Stratiyevskaya don't mention much of anything about appearance in their type descriptions. So visual identification is by no means an established and enforced typing methodology in socionics communities (some polls, like this one
http://personalitycafe.com/socionics-forum/149544-visual-identification-yay-nay.html, show that only 25% of people use any v.i. in typing), rather it's more like minor remarks made from personal observations of a few socionists. Both Viktor Gulenko and A. Aushra were likely SP/SO in another typology they knew nothing about, and people of this enneagram stacking frequently orient themselves by reading various physical features of their environment and other people around them. SP is the most physical instinct, so it's no wonder that their descriptions of types are full of physical details that are of no note or importance to many of others who get into socionics. I usually cross verify if anything written in the profile corresponds to some of the people of that type that I've met, and if it doesn't I simply skip it.