I'm not sure Nintendo will be inclined to create a third (4th if we count the 2nd portable profile) performance profile for Switch.
Indeed. So... keep the 2 mobile profiles the same, and leave the rest to boost for a supposed higher end SKU (memory clocks & docked super switch). Leave it up to the devs if they want to take advantage of higher docked clocks?
The second model can focus on optimized power consumption & cost reduction for costs.
Keep most of the manufactured bits similar enough for bulk orders (chassis, screen, joycons, mainboard).
IMO, literally just cherry picked APU for the higher end model for folks that are looking to use it as more of a docked unit whilst still having the option for mobile (albeit with compromised battery life compared to the lower end). IIRC, there was that data about how Switch owners were utilizing the Switch -> exclusively handheld, exclusively docked, or both, and I think maybe they can make a two-tiered scenario that uses as much redundancy in manufacturing to cater to the two interests without forgoing the idea of the Switch, if that makes sense?
This is problematic for game development, and I am not convinced that even doubling the performance of the Tegra X1 would result in anything more than being a bit closer to the worst performing dedicated console in the Xbox One. I believe a Switch unit that is smaller and doesn't have detachable joy cons is likely.
Tegra X2 is effectively the same HW units as TX1 but with double bus width and higher clocks (they can disable the twin denver2 cores if they don't plough any further R&D into an area-optimized design). Tegra Xavier is the one that has 512 shaders and the oct-core thing.
IMHO, having a separate unswitchable switch just adds extra manufacturing cost in a larger context in the sense that it's a separate line, but I can see the argument for something akin to 2DS. I just think it'd be cheaper to take advantage of existing manufacturing lines for the additional bulk economy of scale there.
But.... I guess we'll see...