I have posted hundreds of measured plots of CD waveguides (and systems incorporating them) extending to 20 kHz and beyond over years of EconoWave development. Of late (this year, particularly), I am also providing polar directivity maps, including some of conventional cone and and dome tweeters.
Compression driver exit and waveguide throat size limits dispersion in the last octave; wide dispersion doesn't happen up there with anything larger than 1". If you want it at 30 - 50 kHz, the throat must be smaller than this, even. It's why Geddes doesn't use anything larger than 1".
Further, depending upon compression driver construction and diaphragm material, there are limits to how high frequency response (on-axis included) can be successfully achieved with HF comp; boost some of them beyond pistonic into breakup mode, and they get nasty real quickly.
JBL 2435HPL with beryllium diaphragm is allegedly pistonic up to 18 kHz. Above there, consensus is that it is best (sonically) just allowed to roll off naturally. Look to JBL's TOTL 3-way consumer offerings using compression drivers for guidance as to the appropriate limitations with this technology. Greg Timbers knows what he's doing with them.
As to Revel's use of domes, see the horizontal directivity waterfall measured by John Atkinson where it's clear that Salon2's exemplary wide controlled dispersion collapses in the VHF:http://www.stereophile.com/floorloud...el/index5.html