(This is a major revision with corrections to the original post. I was much more careful in this revised testing in making sure the laser pointer was exactly perpendicular to the wall and removing the dovetail saddle plate to eliminate its own non-perpendicularity so as to more accurately measure the non-perpendicularity of the RA-Dec axis of the mount by itself.)
This should be my last addition to this subject. I did another orthogonality test by simply using a laser pointer with the dovetail mount removed and my camera to record the path of the laser beam projected perpendicularly to a wall. The short story is that the non-perpendicularity is exactly the same as was calculated with TheSkyX TPoint software model.
I performed three tests, each with the laser pointer mounted one-hundred-twenty degrees from each other, using the saddle mount bolt holes as guides. I then simply moved the mount in both RA and DEC and recorded the laser path. I then measured the perpendicularity using the measure feature in Photoshop 5 Extended version of the two laser light paths and came up with the following results:
All three measurements were in agreement with each other and verified, within the accuracy of measurement, the results reported by TheSkyX TPoint software model. The non-orthagonality was measured to be 0.4 degrees +- 0.1 degree. This is equivalent to 24 arc-minutes plus or minus 6 arc-minutes which is within the range of 26.5 arc-minutes reported by TheSkyX TPoint software model.
Here is one of the photos used to measure the non-perpendicularity:
Bottom line; this particular example of a CGEM telescope mount sucks donkey dookie for the purposes of astrophotography and is basically a non-repairable very expensive worthless Chinese paper weight.
Update added 2012-10-03:
I did a second TPoint calibration run that resulted in a pointing error of 28.6 arc-seconds RMS and using a much closer polar alignment with a total of 6 arc-minutes error. The results for the combined non-perpenducualrity were actually worse at just over 30 arc-minutes, just over one-half a degree or greater than the angular width of the Moon. The sigma figure was less than 50 for the non-perpendicularity indicating an even greater statistical significance from my original calibration run.