fix: languages with combining characters cannot be searched

[SychO9]

Changes proposed in this pull request:
The regular expression used in the Fulltext search gambit attempts to remove non-words to avoid triggering MySQL boolean mode, however, it also removes special characters that combine with words in languages such as Telugu and Devanagari.

This pull request tweaks the regular expression to take those into consideration and allow searching in those languages. (Before this fix, searching for example for: नागरी resulted in actually searching for न गर )

Necessity

• Has the problem that is being solved here been clearly explained?
• If applicable, have various options for solving this problem been considered?
• For core PRs, does this need to be in core, or could it be in an extension?
• Are we willing to maintain this for years / potentially forever?

Confirmed

• Frontend changes: tested on a local Flarum installation.
• Backend changes: tests are green (run composer test).
• Core developer confirmed locally this works as intended.
• Tests have been added, or are not appropriate here.

[SychO9]

for the \p{M} refer to https://www.regular-expressions.info/unicode.html

[askvortsov1]

Wow... This is kind of staggering, not gonna lie: if this has been the main reason completely breaking UTF-8 search, this might genuinely be one of the most impactful PRs in Flarum history.

[askvortsov1]

Only thought: perhaps we should also test that we can search titles for other languages?

[luceos]

At least there's tests now 🙈

[justoverclockl]

this will affect also filter options? because actually filter doesn't work with other languages:

 app.store
      .find('discussions', {
        'filter[q]':  ' తెలుగు భాష',

[SychO9]

when you use filter[q] you're performing a search, so yes.