About frequency offset: I don't know this particular circuit but I assume the reference frequency generated by the GPS engine is an internal clock signal which is programmable. Hence the required 1Hz ('few Hz') offset on the 10 or 40 MHz level.
Afaik the Leo Bodnar unit has the same 'problems' and the solution to offset a tiny bit on the reference level to reduce jitter is known also there.
It all depends on the requirement.. if you need some reference frequency for your counter to display the correct number you go for accuracy. If you need something for narrowband ops you go for stability.
Can it be solved? Sure.. but then you need to correct the OCXO with a GPS engine and run it with really long time constants (1000's of seconds!). Then you would need to shield all of this against temperature variations (1/10 degC is noticable!) which needs a lot of volume and mass.
Can it be solved for the price listed in the AMSAT-DL shop: no. A suitable OCXO itself starts at 100 Eur alone. The OCXO on the board would likely not be suitable for GPS-control due to its instability.
For those who want to know more about this very interesting topic (how to build a GPS stabilised oscillator) I can recommend two sources:
Uli Bangert DF6JB (SK): http://ulrich-bangert.de/AMSAT-Journal.pdf
And from the former AMSAT-NA president Rick Hambly W2GPS: https://gpstime.com/files/TOW/tow-time2015.pdf
73s Achim
PS: and that's why I believe only very few frequency 'measurements' made of the QO-100 beacons.