Here is your problem and how to fix it. I just went through the exact same issue myself when I changed my image train because of new equipment. Nebulosity has given you the big clue and how to find the problem. Your pixel scale is not what you expected. Nor is it what you thought it would be. This is not lazy math, it is just that things are not what they always appear.
That focal reduction amount from your focal reducer assumes a precise spacing from the back of the reducer lens to your camera sensor. If your distance is spot on, you will get the advertised reduction and the expected focal length. However, if your spacing is off, even a little bit, your reduction factor will not be as advertised and your focal length will be different from what you thought it should be.
If the spacing is too close, your reduction will be less than advertised, and your focal length will be more than you thought. If your spacing is too far, your reduction will be more than advertised, and your focal length will be shorter than you thought. This is what is reflected in your astrometry numbers. Even a slightly off focal length can kill your plate solve. (PS2 seems to be a little more forgiving than ASPS)
You can be off a little bit and still get a good focus and reasonably flat stars in the corners. However, if you are even a little bit off, your focal length will be changed due to different than expected focal reduction. If you are OK with the quality of your stars, then don't change it. Just use the "actual" focal length. Again, you can get this from the setting assistant in ASPS and Astrometry. If you want the advertised performance, and the expected focal length, then carefully measure and adjust the distance between the back edge of your reducer glass and your camera sensor.