he is coming to a personal limit of how much he can improve the emulator
That's a bold statement. Do you just guess it? I've read a lot of what Exzap said that past 2 months and he appears to be a Professional-level coder at what he is doing - Cemu is literally his full time job after all - and he has repeatedly shut-down technical statements done by extremely smart people when schooling them, he is not just some guy.
Most people are not aware that Cemu right now is VERY optimized already. It has done ALMOST ALL optimizations that we traditionally see on emulators (well-coded recompilation of game code to native code, multithreading of the main hardware for the most extend, even deep Graphics optimizations that are not even that much needed sometimes).
The improvements he can do right now creep up to the most extremely hard. For instance, move a small part of the CPU emulation operations to the GPU and VRAM (to the extend the simplistic nature of GPUs can handle it), otherwise optimize everything that already exists with minor iterations, or code a Vulkan graphics backend (which might have minor advantages in practice, the PPC CPU emulation thread appears to be the biggest CPU hog after some OpenGL API sniffing I did).
Cemu is a beast, any team would find it hard to optimize it. The ONLY big difference a big team would do is to add easy stuff that Exzap might not bother to do. e.g. Gyroscope emulation, but I suspect Petergov is working on those. The rest of the easy stuff are minor (e.g. who cares about Save states, Nintendo saves are easy nowadays, it's not N64 Zelda anymore).
Otherwise, you can re-write a Nintendo game engine on the fly and make it have pure PC game performance, don't hold your breath that ever happens by anyone.
.
.
