Thanks Michael for added detail and photos.
Process elimination; - I'd be inclined to swap Y output leads at point of connection to tube and try make sense of result.
Also, re your screen shot 475-trace-bent-6.jpeg, in terms of geometry, the image appears to be trapezoid shape. That is, the distance from left to right across the top of the pulses is shorter than across the base line, for the same number of pulses. This is a good clue. Looking at the Horizontal (X) sawtooth linearity between blanking, might offer a reason.
Sorry if you have explained this earlier, but when the Vertical Shift is used to lower the trace, does the top (trapazoid) line width "expand" out such that when the tops of the signal pulses are across the bottom of the tube face, do they now match gratical exactly as they do in that photo?? If so, it reinforces point - the tube geometry is somehow trapezoidal instead of rectangular in a 10:8 ratio.