I'm talking about chips by e.g. Cirrus as used in $5 DACs from China. Any jitter from cable reflections will be above the PLL corner frequency and thus a non-issue.
I see the issue in how you are approaching this and why you may have the belief you have.
You are approaching the jitter as a purely random event. However, reflection induced "jitter", and perhaps that is a poor word to use in this case is more like a clock teetering on a sharp peak. Mainly it stays on the peak, but then it falls off the edge a bit, which could be one clock off, or it could be several in a row, and suddenly you are getting in the range where the jitter attenuation is not that great. Cypress doesn't even spec jitter attenuation out past what, 100KHz, where it is about 10db and it rapid drops to 0 below that frequency.
Now add in some electrical noise from a poor implementation out in the great consumer wild, a touch of power supply noise, some coupling into the signal from EMI caps, even a small bit, and those reflections at inopportune times can do strange things.
The Stereophile article was made in 1993? Old, but we most certainly had SPDIF receivers with PLLs with low corner frequencies back then. The jitter measurements are sub-20KHz in the article.