Property-test Markdown frontmatter parsing

[adewale]

What

• Add a structured Hypothesis property for TOML frontmatter plus arbitrary Markdown bodies.
• Assert exact metadata, body preservation, and body-line accounting through the production parser.
• Add Hypothesis to the locked development dependency group.
• Update the locked Wrangler toolchain to 4.127.1, which carries the fixed Undici required by the repository's high-severity audit gate.
• Record the TOML/Markdown generator and oracle rule in docs/lessons-learned.md.

Why

Example source files combine two grammars at one trust boundary. The existing unittest suite covered examples but did not generate escaping, Unicode, empty values, and arbitrary Markdown combinations with shrinking.

The hosted audit also began rejecting Wrangler 4.114.0 after new high-severity Undici advisories were published. Wrangler 4.127.1 is the first supported release selected by npm's remediation.

How

The generator builds valid TOML string fields and an independent Markdown body, then calls _split_frontmatter and checks both sides of the boundary exactly. The test remains collectable by the repository's unittest runner. The project lesson preserves the need to reach the valid-frontmatter path without rebuilding the production parser as the oracle.

Testing

• Full Python suite: 225 tests passed
• Ruff passed
• Clean npm install succeeded, Wrangler reports 4.127.1, and npm audit --audit-level=high found 0 vulnerabilities
• Independent review reran the generated unittest property
• Full Worker verification remains delegated to CI because the repository requires Node 22 and the local host provides Node 26
• The focused unittest property was rerun after the lesson update: 1 passed

Risk

No application-runtime code changes. The Worker development toolchain moves within Wrangler 4.x and CI exercises it under the repository's required Node 22 runtime.