This right here is why I have serious doubts about the ultimate usefulness of LLMs in coding.

Code is the spec, and it's far faster to write the code yourself (if you know what you want to do and how to write code) than try to pass all this context to an LLM in the form of lossy and imprecise human language.

Yes. Because at the end, the code should be the spec or the final source of truth for it.

I disagree. The spec should be the spec. The code is just an implementation detail.

The style of the code is an implementation detail, once implemented it enters the realm of spec.

Yes, especially so if you do not much care whether it's correct.