Description of prefer-query-selector is misleading WRT “consistency”

[Alhadis]

The description of the prefer-query-selector rule states:

Prefer .querySelectorAll() over .getElementsByClassName() and .getElementsByTagName()
It's better to be consistent.

"Consistency" has nothing to do here. These methods aren't synonymous with each other: .getElementsByTagName() and .getElementsByClassName() return an HTMLCollection object, which updates dynamically as elements are added or removed from the DOM:

const pages = document.getElementsByClassName("page");
console.log(pages.length); // 1

const newPage = Object.assign(document.createElement("div"), {className: "page"});
document.body.appendChild(newPage);
console.log(pages.length) // 2

Conversely, .querySelectorAll() returns a static NodeList which does not reflect changes to the DOM:

const live = document.getElementsByTagName("div");
const dead = document.querySelectorAll("div");
live.length === 3;
dead.length === 3;

document.body.appendChild(document.createElement("div"));
live.length === 4;
dead.length === 3;

The rule — or at least its description — should be amended to acknowledge this.

[sindresorhus]

"Consistency" has nothing to do here.

"Consistency" as in always using the same method that behaves the same.

These methods aren't synonymous with each other:

I'm well aware of how they work.

---

But sure, we can make the difference clearer.

[HERBOD7]

@sindresorhus I can do it.

[feedanal]

This rule basically breaks code that relies on a "live" HTMLCollection returned by .getElementsByTagName() and .getElementsByClassName(). I was under impression that linters are NOT supposed to change code functionality, and prefer-query-selector goes against this idea, maybe I have wrong expectations for linters, dunno.
I can disable it for sure, but why on Earth would someone replace one method with a non-equivalent one??