It has been enabled mainly by the the advent of streamlined tooling to assist with 1:1 byte-by-byte matching decompilations (https://decomp.me/ comes to mind), which allows new projects to get off the ground right away without having to reinvent basic infrastructure for disassembling, recompiling and matching code against the original binary first. The growth of decompilation communities and the introduction of "porting layers" that mimic console SDK APIs but emulate the underlying hardware have also played a role, though porting decompiled code to a modern platform remains very far from trivial.

That said, there is an argument to be made against matching decompilations: while their nature guarantees that they will replicate the exact behavior of the original code, getting them to match often involves fighting the entropy of a 20-to-30-year-old proprietary toolchain, hacks of the "add an empty asm() block exactly here" variety and in some cases fuzzing or even decompiling the compiler itself to better understand how e.g. the linking order is determined. This can be a huge amount of effort that in many cases would be better spent further cleaning up, optimizing and/or documenting the code, particularly if the end goal is to port the game to other platforms.