Preserve required content SEO metadata in workflow

[sergeliatko]

Closes #16

Outcome

• publishes content-production-workflow 1.0.1 with required SEO title and meta-description preservation
• carries separately held values as explicit document metadata while retaining opening Markdown YAML front matter and a complete HTML document head as accepted alternatives
• treats body-only content with unresolved required metadata as incomplete input and forbids H1 inference or invented replacements
• advances catalog metadata to 0.3.1 and prepares the patch package release v0.3.2

Scope

The recipe keeps the same required top-level input, 13 step IDs, 13 evidence-plan entries, capabilities, source and use classes, evidence units and maxima, target, bounds, and result contract. No portable runtime guidance, schema, provider behavior, pricing, or private implementation changes are included.

A registered content-boundary scenario covers required fields, all three accepted metadata sources, no duplicate explicit metadata, incomplete body-only input, and the inference and generation prohibitions. The representative completed handoff now identifies recipe 1.0.1 and records the metadata provenance used by the required analysis.

Validation

• npm.cmd run validate passed: 6 recipes, 3 result contracts, 24 contract fixtures, and 27 registered evaluation cases across 12 segments
• targeted source and generated-catalog assertions passed for the issue requirements and unchanged workflow shape
• npm.cmd run release:verify-assets -- v0.3.2 passed two byte-identical builds for three runtime archives and two catalog artifacts
• npm.cmd run release:preflight -- v0.3.2 passed on release tree SHA-256 9359b62f0f0f8c09b7d1856f7d51524432dcc4260d08b3ca2ca881c1a864b1f0
• generated checksum-manifest SHA-256: 156e42f2bca0680fcc1cb993ece6db5f046d16a9de700c425c3fdadc2cdd7e78

The evaluation registry was structurally validated; no behavioral model-evaluation run or paid/provider call was performed.