research spike - use monogram for parser

[theoephraim]

Experimenting with migrating from Peggy to monogram

Reasoning is that Monogram would generate textmate, tree-sitter, and monarch tooling automatically, meaning it could never be out of sync.

Summary

• make the Monogram parser the default path in @env-spec/parser
• remove old Peggy parser wiring and generation files
• keep the vendored Monogram runtime + strict parser implementation from the spike
• add a parser benchmark script and simplify scripts/devDeps accordingly

Validation

• bun run lint:fix
• bun run --filter @env-spec/parser typecheck
• bun run --filter @env-spec/parser test:ci
• bun run --filter @env-spec/parser bench:parsers
• bun run test:ci (currently fails in existing packages/varlock/src/lib/test/exec.test.ts due to ANSI colorized stdout assertion, unrelated to parser changes)