New PEG parser can parse all our Rune files

[Bill Cox]

I've pushed some CLs that are still in-flight internally, and will push an update once they are all submitted.

This latest update creates a PEG (Parsing Expression Grammar) parser in bootstrap/parsegen/pegparse.rn.  The entire effort took less than a week, demonstrating that Rune has become a productive programming language.

Parsing 348 Rune files runs in under 0.4 seconds on my workstation, and appears to be almost entirely I/O bound, so the PEG parser (with memoization and 1-token lookahead) appears to be fast enough for real-world use.

Next comes building an AST (and possibly a parse tree before that).

Bill