Hmm. I think you wrote the above in a haste.
They do not mean that distant objects shouldn't be rendered smaller than nearby ones. Rather they assume all rays of light coming out from the rear lens element to be parallel (and thus angled exactly 90 degrees to the sensor). As a result there would be no change in focal length when focusing.
I didn't write it, I just quoted it. The article explicitly says "This allows the lens to be focused to different distances without changing the size of the image."
I admit that this is hard for me to conceptualize -- but for a lens to be "focused at different distances," it seems clear that the objects focused
upon must also be at different distances. And if this can be done "without changing the size of the image," then it must be saying that objects at different distances will be imaged at the same size.
Yes, it sounds crazy -- but no crazier than object-space telecentricity, which allows objects at different distances from the lens to be imaged at the same size on the imager, and which is what makes telecentricity useful for optical-measurement applications.
Clearly the photographic effects of this would be weird compared to what we're used to, so I'm sure that true telecentricity works only under very limited conditions. For example, I read elsewhere (but didn't save the link, unfortunately) than an object-space telecentric lens has to have an entrance pupil larger than the field to be measured. In other words, if you want to set up a machine-vision application to measure parts that are nominally 1 cm long, you need a lens with an entrance pupil larger than 1cm.
That's not a big problem at such a small size... but it implies that if you wanted to design an object-space-telecentric lens for photographing people, you'd need a lens more than 2 meters across! (I guess that explains why we can't just go out and buy one, even though it would be hugely useful -- imagine being able to do sports photography, for example, with a lens that showed a player at the same size whether he was on the near side of the field or the far side!)
If we assume that image-side telecentricity is just the inverse of the same principle, then it implies that the lens' exit pupil would need to be larger than the size of the sensor to be covered.
And that, I suspect, is what led the previously-quoted engineer to write that a truly telecentric lens for Micro Four Thirds couldn't be faster than f/2. I don't know exactly how he computed that, but presumably based on the diagonal size of the 4/3 sensor (22.5mm), the lens mount diameter (41.3-something mm, if I remember correctly) and the flange-to-sensor distance (20-ish mm) a lens big enough to have an exit pupil larger than 22.5mm and mounted far enough away to clear the flange distance would have some of its rear diameter masked by the lens mount.
And you're right, for the most part this is just optical-nerd esoterica. I offered it only as an explanation of how an engineer might determine that a TRUE
telecentric lens for Micro Four Thirds couldn't be faster than f/2, while a photographic lens that's only "telecentric" in a marketing sense (i.e., designed along telecentric principles, but not strictly telecentric)
could be faster than f/2.
I can definitely state through practical experience that my 50mm f/1.5 Nokton lens on the G1 lets me use a faster shutter speed when set to f/1.5 than when set to f/2, so it's clear that real-world non-telecentric lenses don't adhere to an f/2 maximum aperture limit.