Ok, I tested with your method and I can see what you're saying, but here's the thing:
Your reference points seem to be skewed to get these results, if I center Mewtwo on the 2nd blue square I don't get a ledge cancel, but your other reference point is moving 100% towards the next square, while moving about 10% of that distance (inside the 2nd square but not perfectly centered) already allows for ledge cancelling.
You are right that you get more distance when running but the distance is so minimal that I actually doubt it's part of Teleport's properties. It also doesn't happen when walking or in the air, you might theorize that it's because the momentum is so low when walking/jumping, but I think it's just that the difference in distance is due to something else entirely.
My conclusion from doing these tests is that Mewtwo's center from a dash is not calculated where the center of the shadow is shown but instead is calculated about where the front corner of the shadow is. This is about the difference you get in distance and it explains why Teleporting backwards when dashing and Teleporting forward seems to give a very slight different result in distance, but it doesn't give any different results while walking or in the air (which both have momentum based too).
So while your conclusion of Teleporting a bit early/late when Teleporting from a dash is correct, I don't think the premise of Teleporting being affected by momentum is correct.