[ENHANCEMENT] - Replace unicode codepoints without regex

[joshmcrae]

This PR refactors the LitEmoji::unicodeToShortcode() function so that known emoji are replaced with their corresponding shortcodes using str_replace. The previous approach involved tokenizing the input string using regex generated from a list of known emoji, but this had several drawbacks:

• A new regular expression had to be generated any time support was added for new emoji
• The regular expression was simplified in such a way that subexpressions matched ranges of characters that made up emoji, leading to non-emoji being discarded, e.g. unicodeToShortcode() removes mdash symbol #36
• The algorithm used to find matches would pick shortcodes for the currently consumed sequence, even if the sequence was part of a longer compound emoji

Because str_replace does not take multibyte character encodings into consideration, all replacement must happen in the encoding of the emoji in the $search argument (in our case UTF-8). For this reason, all functions now convert to UTF-8 internally and return results in the original encoding. These functions will attempt to detect the encoding of the input string automatically and will assume UTF-8 when this fails. A second argument can also be passed to these functions to hint at the input string's encoding, e.g.

LitEmoji::unicodeToShortcode('I like 🍦!', 'UTF-32').

The returned string will be in the original encoding.

In addition to this enhancement, the library now requires PHP 7.4 or greater and has been updated to make use of newer features.