Note that in both your quotes, they're talking about the GPU shaders. Which won't have to be rewritten because most platforms compile shaders at runtime anyway.
I'm not an expert on emulator theory, but what most emulators do what's called dynamic recompilation. It's pretty much the same thing (in the context of emulation) as JIT, sometimes shortened to dynarec. I think the Sony patent had to do with caching the recompiled code, which as far as I understand is a bit more advanced than the average emulator. It's still not magic, and probably has some significant downsides; buggier if it doesn't properly detect slight changes in compiled code, and probably more uneven performance as the dynarec works on new code.
Keep in mind that if the PS4 does use an 8-core Jaguar, it's about half as powerful as a Cell in terms of theoretical peak flops. So even if it were possible to perfectly translate Cell code to PS4 code by machine ahead of time, the Jag still wouldn't be able to keep up. Let's just say that doesn't bode well for any sort of software emulation.
There's two bright spots here. One, realizing that the last-gen CPU is such a monster that it's still relevant for certain media and gaming-related workloads, it starts to make sense to just throw it in and keep using it for both new and old games.
Two, since the Jaguar is 8 cores, the idea of emulating the Cell at a CPU microcode level starts to become possible... well, it would if the Jag ran at double the clock speed or could execute two non-native (and sometimes incredibly complex) instructions per clock. And that's assuming AMD worked some serious magic with the cache controller. So, uh, better hope Sony decided to include the Cell as an integral part of the PS4.