Better spec compliancy for rhemyn

[Rob23oba]

Reworks the rhemyn parser and extends it to additionally support the following:

Regular expressions	Parsed object
^	{ type: 'Begin' }
$	{ type: 'End' }
\b, \B	{ type: 'WordBoundary', negated: boolean }
(?=...), (?!...)	{ type: 'LookAhead', body: [ ... ], negated: boolean }
(?<=...), (?<!...)	{ type: 'LookBehind', body: [ ... ], negated: boolean }
(?<name>...)	{ type: 'Group', body: [ ... ], capture: 'name' }
(?:...)	{ type: 'Group', body: [ ... ] }
(...)	{ type: 'Group', body: [ ... ], capture: true }
...|...	{ type: 'Disjunction', options: [ ..., ... ] }
\1, \2 etc.	{ type: 'Backreference', number: number }
\k<...> etc.	{ type: 'NamedBackreference', name: string }
...{min,max}	{ ..., quantifier: [ min, max ] }
\p{...}, \P{...}	{ type: 'CharacterClassEscape', classType: 'UnicodeProperty', property: string, value: string | null, negated: boolean }
\d, \s, \w	{ type: 'CharacterClassEscape', classType: 'Digit' | 'Whitespace' | 'WordCharacter', negated: boolean }
.	{ type: 'Dot' }

The last two changes have been done to move these options to the compiler and expose these class escapes to the user.

Additionally, there are now better error messages and unicode mode is supported. In fact, everything except for unicode sets mode should be supported now except for a few validation checks.

However: these features have not been implemented in the compiler! The compiler does not account for the new expressions and will throw errors, even for a few things that worked before! This is expected to change in the future.