With the amplitude of the extra components down 40db, that is probably good enough to not cause the problem anymore. I am guessing you could use the old version and cycle through all the multipliers and see at which point it stopped having issues, eg its -6db at 5us, -12db at 15us, etc.. at some point you will see where the clock signals are down low enough to not affect it. If someone has an unpatched scope, you could try and see at which multiple the jitter goes away, and that would allow you to calculate at what signal level (-db) the extra clock harmonic does not affect the scope anymore even before this fix.
And I am guessing that there is NO way to actually get rid of these components totally unless you add extra bandpass filtering, and that would require more LC filters or something and not just a simple ac coupling cap on the PLL output.
More than likely the scope math works fine as long as the level of those extra clock products are down below a decent threshold. I mean -40db is 1/10,000th less signal level than the main peak. That is pretty damn good actually compared the original output where the first product was only -6db down, which means the first product was fully at 25% of signal level of the peak.
So you have gone from the first product being .25 of the primary to only having the power of 0.0001 of the primary signal.
That would be considered extremely good oscillator output quality improvement, in the world of RF where we deal with oscillators, and multiplier products all the time.
I say good on Rigol, that is much better results than I would have expected to achieve without physically adding extra physical bandpass filtering. At those levels, the software should now have zero trouble detecting the primary clock output and rejecting the spurious harmonics.
So, Dave, now is it time to get out that first hour of the review you already have shot?