Most of cases phase noise is measured in dBc, so a simple balun placed at the differential output will help to do the measurement using a spectrum analyzer.
agree with that one, a balun will not add phase noise. A differential amp will add noise.
Just make sure the output power of the Balun is high enough to satisfy the measurement systems input power requirement for the noise floor you are trying to measure. If it is not too tight a spec., you could add an amplifier after the balun to boost up the power.
The balun will reject noise that is common to the two traces, just leaving the sum of the differential noise. I think it would be difficult to infer from the single ended output how much common mode noise may have been on the traces. However, as long as the LO is going to a differential device, it will also reject common mode noise, so from the system perspective the differential noise is probably most relevant.
I see that I have to underline my previous post.
Balun works better, and add less problems than any other topology when want to go from differential to single ended for Phase Noise measurements.