As I understand it, and as you yourself are writing, the GSDO is very stable over longer time, several minutes to hours and even days. The Rubidium should be perfect for the short time readings, frequencies in a minute or so.
The reason for a 1pps GPSDO should be to correct the Rubidium every second, while the Rubidium keeps a more precise timing between than a GPSDO would ever be able to.
Yes I know 12 digits is extreme and rarely usable, but I hope to build some crystal oscillated clocks that shall run independent and fairly precise, so to be able to test them, would it be great to have something that shows exactly what deviation a crystal have, without having to let them run for days or weeks.