And not the final word... but more to think about.
Even with densification, not all things can be treated as equal 
well Duh. I didn't test putting projected data in the map. Seems very obvious now that I know this was the "glitch".
It seems like a big change from Desktop, to not be able to rely on projecting on the fly and 'borrowing' a projection from the data frame. But with your explanation of extent densification, I see why Esri is trying to control or constrain the trouble many blindly get into.
And about your comment on web mercator auxilliary sphere! Most of my instruction about CSs is to demonstrate how ridiculous measurements are if left in Web Mercator. Esri has taken two major lab exercises off the table with this change. 
At first I was resistant to the idea of HAVING to reproject data because I've always been told there is a small erosion in the precision of the data each time you permanently change projection. I thought projecting on the fly was a great way to use display coordinate systems to supply the best CS for the analysis or measurement. But maybe I am simply resistant to change. And also working from a much more rudimentary place than others like yourself.
I am coming at this from the perspective of an instructor, not from the research or hard core analysis side. I am trying to demonstrate some of these basic concepts with CSs. I'll just have to get more creative.
I'll be reading up on densification. Thanks a lot for this conversation, Dan. I've never used GeoNet before and found this to be incredibly helpful.
Good luck Shannon. PS, I taught University level GIS for 28 of my 33 years... you will never stop learning