Yeah. And same as with the via discontinuity, where traces cross a plane gap, the image current spreads out under the plane and around the gap. If the gapped plane happens to lie over another (suitably solid) plane, then there's a path for that image current to be carried across the gap, and it doesn't look any worse than a via.
Which is also your observation -- you've posted TDRs of both cases.
Even if you remove one plane entirely, that just increases the height to the next. Say, for a trace on the top layer, with removed 2nd layer plane, and either 3rd layer or bottom layer ground.
The extra height is only increasing the trace impedance, and giving the fields more space to spread out in. Which increases the amount of trace-to-trace clearance you need to avoid some amount of crosstalk, but that's all just geometry scaling, nothing crazy.
Traces without any material nearby, just a straight hole through the board, are where you invite problems, and the only case you really need to worry about.
Yup. 1.6mm board looks like, we'll say 2mm wavelength because of c, but that's 1/4 wave, so 8mm, which is 26ps. 100ps is 4x that, so you'd just begin to see the effects. In the 50-100ps range is where you'd be concerned about optimizing via diameter (id/od) and clearance.
Hmm, you could also neck down the trace slightly, say for 0.8mm before the via pad, to add some inductance there. Now it looks like a LCLCL filter: more poles, more degrees of freedom to optimize, flatter bandpass into a sharper cutoff!
An amusing mental exercise, and delightfully useless. You'd give up FR4 long before anything this precise was a concern!
Tim