Today, I talked with my colleagues on the position of the photo-diode.  In short, measuring the reflection is a bad idea as opposing planes in practice are not planar due to fabrication errors. I would therefore not correct for the jitter. The idea now is to add a prism or mirror and deflect the laserdiode beam upwards before it refracts through the lens or after it refracts through the lens.
This prism or mirror is imaged to be small, e.g. a cube of 5x5x5 mm. The light is hereafter measured by a photodiode. This measurement would account for the aberration. In my initial design, I used a mirror. The disadvantage is that it is harder to mount at an angle of 45 degrees. The prism has the 45 degree angle build in.