Well, I have to agree with Osiris here - he is being *rightfully* sceptical (because it does not *seem* to work for him), and discussion is rather civil compared to BROL to be fair.
After all, Jim explicitly ASKED for a scientific analysis of his claims, and Osiris is doing just that, and for free!
If he is getting result you don't like... well, don't shoot the messenger, unless you can prove that he scrambled the message (which is also possible, to be fair).
And since Zwift does not have any way to 'prove' that you are riding a recumbent, it is rather irrelevant for proving anything about them (though makes an excellent training platform).
As for 'magic of the front triangle' - after some discussions with frame builders I can say that those claims must be true, but you must compare Vendetta not with carbon wonderbikes, but with more pedestrian aluminium bents - and having triangulated drivetrain should indeed allow more (maybe MUCH more depending on your pedalling style, power output and frame design) efficiency in power delivery all else being equal, or much ligher weight, or both.
To prevent chainstays bending from chain torque and, hence, rear wheel minutely misaligning from direction of travel and related trail-like friction forces you need either VERY stiff chainstays (in case of RWD bents they are 'one large chainstay') and/or triangulated drivetrain.
Plus, like you said, you must certainly get about 3-4% extra power from reduced friction losses in idlers. In case of multiple idlers and chain tubes losses might be even more significant.