FSR, by design, is hardware agnostic, designed for maximum cross platform support, hence integration from consoles to phones to PCs. It is not like XeSS that has three different kernels, or DLSS, that requires tensor cores and is specifically built around NVIDIA architecture. FSR3 isn't really changed from FSR2, it just adds some features for attempted parity with NVIDIA, such as frame generation.
AMD frame generation technology also does not use or require hardware exclusive to AMD, beyond simply having enough GPU horsepower to generate a good enough base FPS, unlike DLSS3 FG, which requires the optical flow accelerator within Ada cards, so there's no real difference there either.
Your argument would be more suitable to instead support also running game testing on Intel Arc GPUs, but without speaking for reviewers, I'd wager it's a limited enough consumer base to not warrant the expense and time at this point, beyond acknowledging that IQ and performance will be slightly better on native Intel hardware. Perhaps when Battlemage comes out.
Besides, other reviewers and the general public encounter similar issues with FSR. Unfortunately AMD has simply not improved the upscaler component, for quite a long time at this point.
Personally I think the XeSS approach has the best balance, DLSS for sure has the best quality, but it's not universal, requiring specific hardware, and FSR goes too far in support, without enough development to make the end result good enough.