Very interesting speaker.
The concern with having two tweeters (and two midranges) side-by-side is that, for laterally off-axis listening positions where the difference in the two distances (distance to one tweeter minus the distance to the other tweeter) is in the same ballpark as the wavelength, comb filtering should be evident, due to the alternating cancellation and reinforcement of the two sources, as wavelength increases (or decreases). The frequency region where this would be most visible would start with wavelength slightly shorter than the horizontal spacing of the two tweeters (also the two midrange drivers) and continue upwards from there until the directivity of the tweeter is such that you have two lobes that don't overlap. And of course the further off-axis you are, the more pronounced it will be.
Given that the two midranges are directed outward and that they become fairly directional by the time you reach the crossover frequency, it is not surprising that you don't see this effect with them. The region where I would most expect to see this effect is in the mid-treble and fairly far off-axis horizontally. You do in fact see some of this effect in the Ronald McDonald directivity graph, from about 3 kHz on up to about 10 kHz. In this graph, you clearly see response peaks at about 4.3 kHz, 5.7 kHz, and 7.4 kHz, with nulls in between. Oddly, though, the ripple effect is nearly as strong on-axis or very slightly off-axis as it is very far off-axis. You should be able to get at least +/- 10 degrees horizontally off axis before you see any of this effect at all. Moreover, the difference in the two distances changes as the off-axis angle changes, especially when you are just far enough off-axis for the effect to begin. As such, the frequency for a peak or for a null in the ripple should change as you move further off-axis, which means that those lines should be curved such that the graph would resemble the sketch of a Christmas tree laying on its side, and not look nearly as ruler-straight as they are. The only explanation I think of is that the angled baffle is doing this, but I don't think this is a good explanation, because I still don't think the lines you see there should be ruler-straight.
The first two graphs indicate that the on-axis response is exceptionally flat, except for what is going on in the upper bass and lower midrange. Up around 5 kHz you do see a swing in the response at 20 degrees off axis, an up-and-down with a difference of 3 dB to 4 dB. This would likely produce some slight coloration for someone sitting 10 or more degrees off axis, but it would be slight and possibly not distracting. The most surprising thing I see is the strong directivity at high frequency. I would have thought that the two tweeters being aimed as they are would have yielded a broader dispersion pattern, but evidently not. The tweeters themselves must have unusually strong directivity throughout their response range. I expect that this is an attribute of this type of tweeter generally, ring radiators without the center dome.
As for that little messy area in the upper bass and lower midrange, this is most likely caused by the protrusion of the baffle just above the upper woofer. That messy business in the upper bass and lower midrange isn't especially bad, but it is probably audible and it is probably the biggest drawback of the design approach.
Which brings the question of whether the SDA actually works. I suspect that if it works, it will rely on early reflections from the side wall. The reflection from the right side wall, for example, would be heard strongly from the right ear and not the left, and since the driver that is aimed mostly at the right side wall emits the inverted left-channel signal, the effect would be to cancel much of the left-channel signal that the right ear hears. What concerns me, though, is that in order for this cancellation effect to work, the left-channel signal arriving via right-side-wall reflection needs to be out-of-phase with the primary left-channel signal coming directly from the left speaker. As such, it seems to me that in order for it to work as it is expected to work, the positioning of the speakers would have to be very particularly coordinated with the width of the room, and you would need to be sitting in the sweet spot. If you don't get this just right, then to me it seems like you could end up with an effect that isn't anything like the intended effect.
It would certainly be a fun speaker to play with on rainy days. But for me personally, it seems like you're paying a whole lot of money for something that isn't necessarily much more than a gimmick. When you're as old as I am, you've seen a lot of gimmicks come and go over the years.