As I recall, peak gain was somewhere between 1/2 and 5/8 wavelengths apart, but I could be wrong. I do know there is an upper limit of distance between said current nodes, and I am thinking that 9/10 (i.e. 0.9) seems a bit high. I could be wrong as it has been a while since I've modeled this stuff...
In a 1/2 over 1/4 or 1/2 over 1/2 scenario, you have a physical 1/2 wavelength of antenna (1/4 on the top and 1/4 on the bottom elements) plus the physical size of the phasing mechanism between the two, of which a coil based system discussed in the G-Max (and in your Cebic link) will normally require more of said space that other methods require (depending on the specific design of course)...
All this being said, I do recall the EDZ models comparing very well against collinear models, which makes me question my thoughts on said distance requirements mentioned above...
Unfortunately as I have been bouncing around between linux distros lately I don't have any modeling program loaded, and the one I am currently running wouldn't be the best to load one up in (its an atomic version and will have trouble getting a version of wine going, for anyone who knows what that means...). It would be nice to get a third party model for comparison to the manufacturers...
Any who, on my end I'm calling my data speculation at the moment, as I'm pulling from memory from years ago...
I see how it is, sneaking in a post after I start typing one of my own 
I agree with this, more or less. However, I think that people can be creative, and can likely get near as makes no difference if they choose to put out the effort (and maybe money?) even on a band like 11 meters...
The DB
