measurement source impedance
The argument here seems to be skirting the basic question: do you need large-signal or small-signal measurements?
It depends. If it is a low power source, with buffer amp after the oscillator, probably a small-signal measurement suffices. Just be careful, as stated, to not overload the VNA front end! Obviously, just putting a 10+ db attenuator between your source and the VNA is not the way to go if you are trying to make accurate s11 measurements.
If you are measuring an oscillator without a buffer amp (very sensitive to load pull), or an output amp that is putting out a lot of power (somewhat sensitive to load pull), using a small-signal measurement will give big errors. It would get you in the ballpark, however, so that you could tune up the result. A formal large signal meaurement, such as a load pull measurement, would be better in this case.
In SOME types of oscillators (GUNN DIODE, for instance), I do not think you could even make the measurement at all in small signal, as the oscillator might jump into the wrong mode during the measurment, and work in a different mode in the final matched circuit!