Refactor Lexer
The lexer needed some cleanup, I found myself doing this as part of a Unicode RFC, but factoring all that out to make the Unicode RFC PR easier to follow.
* Always use hexadecimal form for code values.
* Remove use of `isNaN` for checking source over-reads.
* Defines `isSourceCharacter`
* Add more documentation and comments, also replaces regex with lexical grammar
* Simplifies error messages
* Adds additional tests
