Support Unicode-aware regex

[Scott-Guest]

Closes #4357

Convert K Unicode-based regex to an equivalent Flex byte-based regex:

• Outside of a character class, just parenthesize to keep the bytes grouped
  • r"😊*" becomes r"(\xF0\x9F\x98\x8A)*"
• Inside a non-negated character class, factor any single Unicode character out into an explicit |
  • r"[a😊b]" becomes r"(\xF0\x9F\x98\x8A)|[ab]"
• In all other cases (character ranges and negated character classes), report an error if there are non-ASCII characters

Additionally,

• Check that character ranges [c1-c2] have codepoint(c1) <= codepoint(c2)
• Check that numeric ranges r{n,m} have n <= m

The commit history is incremental, and I'd recommend reviewing commit-by-commit.

[Baltoli]

There's a great line in the documentation for LinkedHashSet:

This implementation spares its clients from the unspecified, generally chaotic ordering provided by HashSet

[Baltoli]

Very nice; seems like a really well thought through implementation!