Sidecar rollout: per-source thumbnails + license + is_public fields

[rdhyee]

Context

Raymond endorsed the sidecar enrichment pattern on 2026-04-17: per-source parquet sidecars keyed by pid, LEFT-JOINed into the wide parquet at build time. This keeps source-specific enrichment out of the canonical PQG pipeline while making it queryable from a single wide table.

Unlike a thumbnail-only sidecar, the schema should carry richer fields that the Explorer and downstream consumers (e.g., Charismatic samples audit #130) actually need.

Sidecar schema (minimum)

Field	Type	Notes
pid	string	Join key
thumbnail_url	string	Image URL for Explorer card & homepage showcase
license	string	SPDX or source-specific license identifier
is_public	bool	Whether record can be surfaced publicly (embargo handling)
media_url	string	Full-res media, if distinct from thumbnail
harvested_at	timestamp	When sidecar was generated (provenance)

Priority order (per Raymond)

1. Smithsonian IPT (partner-first — relationship-driven priority)
2. GEOME
3. SESAR
4. OpenContext — already done (reference implementation)

Acceptance

• Smithsonian IPT sidecar generated and LEFT-JOINed into wide parquet build
• GEOME sidecar generated and joined
• SESAR sidecar generated and joined
• Build script documents sidecar regeneration cadence
• thumbnail_url populated for all four charismatic samples (unblocks Showcase samples: make the 4 front-page images locate themselves on the globe #130)

Downstream consumers

• Showcase samples: make the 4 front-page images locate themselves on the globe #130 — Showcase samples: make the 4 front-page images locate themselves on the globe
• Data pipeline: thumbnail_url column is empty across all 6.7M samples #131 — Data pipeline: thumbnail_url column is empty across all 6.7M samples
• Interactive Explorer sample card thumbnail rendering
• QUERY_SPEC.md thumbnail provenance question (see forthcoming QUERY_SPEC v0.2 issue)

Related

• Landscape draft issue: https://github.com/orgs/isamplesorg/projects/7
• Layered parquet design: Layered parquet: base + ordered overlays for supplements and overrides #135