Begin re-implementing tokenization using classification

[lex] detect preprocessor directives

[lex/simple] Support '+-' in numeric constants

Improve the user interface a bit

[lex/simple] Implement pretty much everything

The lexer successfully parses large C and C++ files, but the resulting tokens
have not been checked, and thorough fuzz-testing is still pending.  It focuses
on C23 conformance.  While escaped newlines (which are truly terrible) are
supported, trigraphs are not, thankfully.

[lex] Setup simple lexing

The lexer correctly implements translation phase 2.  The next step is to perform
the actual tokenization work (phase 3).

Set up stub 'main' and a 'lex' module

Start over with Rust

Add a stub 'lex/avx2.h'

Use 'src' for header files

Add a stub 'lex.h'

Set up EditorConfig

[README] State current development progress

Add a README and a LICENSE

Set up Git