We are on the same page now. I agree, it is a form of backlash, but as I understand it, it is perpendicular to the plane of the blade and direction of movement of the blade. It may be that whatever rides in whatever groove or slot is cocked or forced one way (apex to the right and bottom of the blade to the left) when the blade is driven up, but when it is lowered the mechanism and hence the disk of the blade are cocked in the opposite direction.
I has been so long since I assembled my saw after removing the top and lifting it up to my shop I don't remember what the whole trunnion details look like anymore. But sounds like you have definite slop there. I don't know if I have ever heard of an adjustment for that.
The measurement change could happen with a tilted blade/arbor if the arbor that can move axially in the bearings. I don't think that is the problem because there is no reason for axial motion of the arbor with only the height of the blade being changed.
Did you watch the blade tilt when adjusting the height? Even though you are only adjusting the height, if the tilt mechanism has slop it could be tilting on you without you noticing. You would need a Wixey or other tilt box.
I'm about out of ideas without tearing into my saw to give the trunnion a close look.
Here is a situation I found on the web that would cause your problem:
"Here's the latest update. I took the saw apart and I found that there was
significant play in the arbor assembly. Not the arbor or arbor bearings,
but in the hinge that attaches the arbor flange to the trunnion. When I
apply light pressure to the blade I can clearly see the flange move. The
saw is going back to Grizzly next week. Hopefully, they will be able to
fix it, or exchange the saw."