docs: align skill schemas with intuition-data-structures taxonomy (ENG-9713)

[jonathanprozzi]

Summary

• Adds Data Structures Alignment section to reference/schemas.md teaching agents the two-layer model (classification + enrichment) from the intuition-data-structures repo
• Expands Schema Type Selection from 3 categories to 6 (Person, Organization, Software, Predicate/Concept, Blockchain Address, default Thing) with future classification type annotations
• Adds Classification Mapping Table (21 rows) showing how 35 data-structures types map to current pinThing/pinPerson/pinOrganization mutations
• Adds Field Population Guidance table distinguishing identity fields (name, url, email, identifier) from enrichment fields (description, image)
• References classification taxonomy in SKILL.md Protocol Model section

Framing Decision

Educational + keep populating. Agents learn the two-layer direction but are explicitly told to keep populating all pin fields with the best data available today. No pin mutation signatures, encoding patterns, validation rules, or failure handling changed.

Test plan

• Read through updated schemas.md — classification mapping table accurate against data-structures repo
• Verify current pin mutation docs are unmodified (lines 99+ in schemas.md)
• Confirm field population guidance is practical and non-contradictory
• Test with an agent: "What type should I pin a GitHub project as?" should yield Thing (future: SoftwareSourceCode)

🤖 Generated with Claude Code

[linear]

ENG-9713 Utilize (and test) the new semantic engineering schema structure