Add proper Korean romanization

[alexdraconian]

Replacing #4182, since PR has been compromised by anti-ransomware mass-deleted and re-added files on my environment. Here's some important points.

---

Currently, Korean romanization does not work properly, since romanize() in Clean.php does not handle full character. It only works when we type each component individually(ex. ㅌㅔㅅㅡㅌㅡ), which is virtually useless.

So I added function for decomposing Korean characters and romanize them accordingly.

However, here's the catch.

• This code may have performance impact, since it uses looping instead of strtr.
• However, if I add individual full characters to table instead, 11,172 characters should be added, which is quite a lot.

I thought this PR needs some discussion or review, so I opened as draft. (This code works though.)

---

I implemented dedicated Korean test, with most frequently used words provided by National Institute of the Korean Language.

This implementation does spelling-based romanization, which is not official romanization of Korean(which uses pronounciation-based). However, pronounciation-based romanization is much more complicated and I think this relatively simple implementation works fine for the purpose.

As far as I know, other project(OpenProject) also uses this kind of romanization, but I can't surely confirm.