New rule: Disallow characters which are made with multiple code points in character class syntax (no-misleading-character-class)

[feross]

https://eslint.org/docs/rules/no-misleading-character-class

This prevents a really tricky Unicode regex edge case that always results in a bug. Let's include it in standard 13. Part of the eslint recommended set.

0% ecosystem impact, so it's a no brainer.

[joshgoebel]

Is there some other name I could google to read up more on this I'm really failing to understand this rule.

> /[\u0300–\u036F]/u.exec("\u0302")
null
> /[\u2020–\u2027]/u.exec("\u2026")
null

The linter will flag the former, but not the latter, yet neither seem to work as I'd hope. So I'm wondering if there are some cases this rule doesn't catch but are also problematic?