regex-utils: Add support for translating regex character sets into wildcards when possible.

[Bill-hbrhbr]

Description

Implements the following functionality:

• Reduces a character set into a single character if possible.
  • A trivial character set containing a single character or a single escaped metacharacter.
    • E.g. [a] into a, [\^] into ^
  • If the case_insensitive_wildcard config is turned on, the translator can also reduce the
    case-insensitive style character set patterns into a single lowercase character:
    • E.g. [aA] into a, [Bb] into b, [xX][Yy][zZ] into xyz

Validation performed

Verified in unit tests

[LinZhihao-723]

Before going to the detailed review, I had some comments about the way of tracking the charset. In the current implementation, we have an iterator in the TranslatorState and expose APIs to update/query the underlying iterator that points to the beginning of the charset. However, it's a little confusing and unsafe:

1. The constructor also takes an iterator, which doesn't make sense if it is not the beginning of the charset.
2. The names of m_it, get_marked_iterator, and mark_iterator don't really indicate their usage.
3. When querying an iterator, we don't have a sanity check of whether it is set to the beginning of a charset properly.

I would propose a solution which we can add a dedicated class to track the status of charset escape. It should have a counter of chars processed (indicating the length of the charset), a flag indicating whether it's an escape char or not, and also the storage of the character to proceed. The TranslatorState should have a std::optional member of such a charset status tracker, and emplace the optional tracker when [ is found and reset the optional tracker when ] is found. All the checks can be passed down to the charset tracker's member function to proceed when you see the character in charset_state_transition, and you don't need the chaining if-else with the length check. Defininig sub-states inside this charset tracker class might be useful. I think that would make the interface more clean and readable.