040 literate annotations

// TEST: Literate Annotations
// STATUS: ASPIRATIONAL
//
// Annotation blocks as micro-documents: markdown prose + directives.
// Parser extracts `-` prefixed lines as directives, ignores prose.
//
// This brings literate programming to metadata - the annotation
// explains AND instructs in one place.

~[
# Token Grammar

Defines a lexer for simple arithmetic expressions. Each branch
represents a token type with its regex pattern embedded in the
phantom type annotation.

- derive(parser)

## Generated Events

The parser transform will generate:
- `token.parse` - stream -> token branch
- `token.serialize` - token -> bytes

## Performance

- optimize(level: 3)
- inline

## Whitespace

We skip whitespace tokens rather than emitting them.
This keeps the token stream clean for the parser.

- whitespace(skip)
] tor token {}
| number u64[\d+]
| string string["[^"]*"]
| identifier string[[a-zA-Z_][a-zA-Z0-9_]*]
| lparen void[(]
| rparen void[)]
| plus void[+]
| star void[*]
| whitespace void[\s+]