Good point. However, teaching itself as a skill becomes less and less important the further “up the chain” one goes.
Consider the soft skills needed to lead a graduate seminar, where you have a professor and three to twelve grad students (or upper division undergraduates) around a conference table for 3 hours once a week. Well done, there is still some flow of information from prof to students, but for the most part the format is a meeting where the students have done their homework on the topic at hand, and are prepared to discuss it intelligently. Yes, the prof needs to think through a topic for each of the 10 or 12 sessions in a quarter/semester, and needs to be versed in the field enough to suggest the right readings as prep, but the students are in that room as participants rather than objects of teaching.
Consider also the skills needed to supervise a PhD candidate at a research university. Yes, there’s a whopping amount of learning the norms of behavior in academia that have to be learned somehow, but to be honest a professor who’s at the leading edge in their field, good at grantsmanship, and a “good enough” coach is a better PhD adviser than the best teacher who’s fallen behind. (I went to industry after my M.S. and never looked back; my M.S. thesis adviser was mediocre lecturing in an undergraduate classroom but very good in smaller groups and 1 on 1, and really knew his narrow field.)
Edit: One last thought: at a non-research-centered junior college, college, or university I would expect a greater focus on teaching, particularly teaching lower division material, and much less focus on narrow subject matter expertise. It’s putting those narrow experts to work teaching material they haven’t thought about in years or decades where we get in trouble…