For the sake of argument, let's stipulate that wear occurs only as you describe, resulting in a wear arc offset from the centerline of the chain.
1) In one sense it doesn't matter. At least a major component of the wear is along the chain. Wear (24 pin net wear) is measured along a straight chain. Flip the chain over and the measured wear is identical because nothing has changed. Unless you come up with some other way to monitor wear, you're pedaling blind in terms of the actual condition of your chain.
2) Again accepting your wear mechanism, let's look a little closer at the numbers. On a 40T sprocket (average front ring), the link flexes 9°. If wear occurs evenly along the sweep, the wear arc is offset 4.5°. On a 16T (average rear) sprocket, flex is 22.5° with the arc centered at 11.25°. Tension is the same so the wear footprints are additive and actual arc centerline is roughly 8° offset.
Tension is higher on smaller cogs because there are less teeth to distribute the load over. Tension is also higher with smaller chainrings because you have more leverage than with a larger ring. The actual point of highest wear is probably further than you estimate, but it would depend on a number of factors.
The difference in wear between max wear (at 8°) and wear on the chain centerline depends on the radii of the two wear surfaces. Since the pin-to-inner plate clearance is pretty small, the difference between radii is also pretty small, resulting in a wide arc. Empirically (i.e. from memory), the arc is about 90% on a worn chain. The difference in wear over 1/11th (8°/90°) of that arc is going to be very small - on the order of 1/11 of the clearance. Or if you want to project to the new wear arc, it would be roughly twice that (16°/90°) or 2/11ths of the clearance.
again, you are probably overestimating this effect. The overlap is probably less, and the contact arc is also prbably less than 90 degrees.
In other words, your net gain in wear is at most 18%. Hardly the "double your wear" claimed above.
So you agree with me that the effect is there, you just have a different estimate of its value.
em