unsound - I have remembered that I also tried 'your' configuration in my Middle Crossing studio where we had a fairly large, well-proportioned, well tweaked room. I believe that duramax has two pair of CS5is, but I'm not aware that he has done this experiment. I had two pairs of CS2.2s, back to back and unitized, each channel driven by a stereo Classé DR9, one channel driving each speaker, with another identical amp driving the other two back to back speakers.
Side-step to Thiel's conceptual model - to have each channel act as much as possible like an omni-directional microphone. That causes extra trouble for set-up because the wide dispersion interacts more with the room than narrow dispersion designs. Adding the back-firing speaker exacerbates that set of problems.
On the other hand, the psycho-acoustics favor the 2-speaker arrangement since they are driving the room (as seen and aurally interpreted by the listener) more like real 3D instruments in that room, more like that mythical omni mic. This configuration will add more power to the sides and back than to the front-firing energy. So, more space and/or critical wall treatment will be needed.
My room was 16' wide x 24' long with porous-resistive walls that acted like a much larger room. In that room my standard spacing put 5' from the front (solid) wall to the tweeter plates. Since the back wall was porous, I could pull the speakers into the room without bounce problems from the back wall. I recall bringing them forward to around 9'. I lacked room to spread them wider than their 9' ctr to ctr. That tuning operation is a combination of physical experimentation and experientially gathered positioning data from decades of setup work. Tedium that pays dividends.
Regarding bi-wiring. I put considerable effort into that proposition due to its popularity in the field. I dislike it, like Jim disliked it. In Thiel's coherent milieu, some bonded point must be assumed as all the circuitry and drivers act as a unified system from top to bottom. The only point that we get full control of that unified system is with all drivers driven from a single point of entry. I also found that in the real world of cost-benefit, a 'better' set of cables provides more value than two 'lesser' sets. But even if you could get two identical 'better' sets, their working parameters would be different than those of the speaker as designed due to their considerably longer length and reflection parameters. My best results (half-century and counting) have been with single runs from a stereo channel (not bridged) through a single pair of speaker cables. By the way, in those trials my best bi-wire results were with internally bi-wire in the same cable rather than separate home runs between amp and speaker.