[6.x] Fix rate limiting unicode issue

[driesvints]

This PR fixes an issue where rate limiting would be reset for characters with unicode in them. By converting these characters to regular ones, the same key is re-used and rate limiting is properly applied, albeit a bit stricter.

[oliworx]

This fix will not work as intendet with most of the unicode characters.
The test covers only one simple example from a roman language.

Lets assume a key with CJK characters like 'ユニコード' :

>>> $key = 'ユニコード';
>>> preg_replace('/&([a-z])[a-z]+;/i', '$1', htmlentities($key));
=> "ユニコード"

So the unicode string is not modified at all.

[driesvints]

@oliworx that's not an issue. This fix was purely focused on alphabetic characters which were non-unique for a DB store but unique for the limiter. Other unicode characters have no direct alphabetic replacements and are always unique.