Code is written with two kinds of readers in mind: the computer and other humans. Prose is written just for humans.
With code, there is a very precise set of rules determining whether the code will be readable by a computer. There is no correspondingly precise set of rules for understandability by humans.
I can see the analogy between PEP 8 and grammar rules, but I think there's an important distinction in the provenance of those rules. PEP 8 is a formal proposal adopted by the BDFL. There is no mechanism for submitting or approving formal proposals for English grammar rules. So the spoken language is always the authority on what's correct and what isn't.
Yes, but to some extent that's a peculiarity of English: https://en.wikipedia.org/wiki/List_of_language_regulators.
I don't think the fact that code is also meant to be read by a computer is relevant in this context. But let's take another example if that doesn't work for you: the novel Trainspotting by Irvine Welsh. It's written in a phonetic Edinburgh style. By 'normal' english standards, it doesn't make any sense. If someone was to write in the same style way without any knowledge of English grammar and spelling, you would just assume they couldn't spell. But that's why, in the context of a novel written by a talented author, it's actually a great bit of prose. I could give thousands of other examples.