Your note struck a real chord in me. I too have seen amateurs struggle with collimating their rotating upper ends with the result that they never quite achieve perfect collimation.
Not only does the rotational plane of the upper end have to be exactly perpendicular to the optical axis, but the center of rotation must correspond to the center of the optical beam. For instance, this means that touching a collimating screw on the back of the mirror mount is a no-no: the upper end will no longer be perpendicular to the optical axis and the center of rotation of the upper end will no longer coincide with the optical center. One must instead collimate to the mechanical center of the upper end. This means adding control to precisely shift the primary laterally, to name just one extra collimating step.
Unfortunately, even in a friend's precision machined metal and fiberglass 14" f/5, with a beautiful machined collar that the upper end rode on, we were unable to achieve more than what I would call casual collimation. Just too many things sagged a tiny tiny bit here and there; just too many things to juggle while adjusting; and just too many accidently bumps in use that would cause collimation to go downhill, resulting in another multi-hour session to attempt re-collimation.
My solution is to slant the focuser at a permanently set slight angle above horizontal (this is in a dob setting) for comfortable access when the tube is pointed horizontal and when the tube is pointed vertical.
Mel