At the risk of dragging this on further (this whole thread seems to be getting drawn out), if the laser is properly aligned its axis will be in angular alignment with its mechanical axis. Hence, its axis will be parallel to the travel axis of the focuser.
Therefore, when racking the focuser in/out the beam will travel along its axis. So, the spot location will not change at the diagonal (or anywhere).
Splitting Hairs a Bit:
This brings up a side point. I don't know what tolerance the commercial laser collimator collimators are aligned to. For the one I built, I aligned it by setting it into a vee block and rotating it - just adjust it until the spot doesn't move. To increase the baseline and make it a one-man operation, I doubled the beam back by reflecting it off of an old diagonal mirror onto a white card adjacent to the collimator. I was able to get the beams axis to be parallel to the mechanical axis to within about 30 arcsec (about 1/8" wobble over a 75 ft baseline). Though I am not sure whether it is an apples to apples comparison, for an autocollimator to match this, its reflective surface needs to be square to its mechanical surface to within about 0.00015 inches in order to match this.
I am not sure what accuracy is required for the errors to even be visible at the eyepiece (esp. given the other tolerances in the system - eyepiece element centering/squareness, drawtube clearance/angling, etc).
I seem to remember that the collimating booklet sold by Tectron discusses this - I will have to review it tonight.
Bob