P2 pilot: add the pandas_panel trio (realwage, countries, employ) with manifests

[mmcky]

P2 pilot — cross-series shared static (PLAN Phase 8)

The second pilot: three files consumed by two lecture series (lecture-python-programming and lecture-python.myst), migrated from each series' own-repo copy to a single canonical copy here. This is the flat-namespace-with-two-consumers case, and it retires the last of the pandas_panel legacy-repo references.

What lands here

File	Source	Class	Licence	redistribution
realwage.csv	OECD real minimum wages	constructed (builder unrecovered)	CC BY 4.0	permitted
employ.csv	Eurostat employment	constructed (builder unrecovered)	Reuse w/ attribution (Decision 2011/833/EU)	permitted
countries.csv	WorldData.info country table	verbatim	© WorldData.info, all rights reserved	restricted

Integrity (stands in for CI — this repo has none)

Each file is byte-identical across both consuming repos (git blob compare, 2026-07-17), and each manifest's integrity.sha256 matches the committed bytes — so the repoints cannot change lecture output, and the go-live consumed-file invariant holds. Upstream provenance is unverifiable for all three (inherited, undated, no builder committed) — recorded honestly per the P1 pattern rather than faked.

The countries.csv licence wrinkle

WorldData.info is all-rights-reserved — a no-redistribution term that AGENTS.md would normally block. It is cached here as an inherited exposure: the lectures have served this exact file publicly for years, so caching it here adds no new exposure, while hiding it would obscure a real one. It is recorded honestly (redistribution: restricted + note) and logged for licence review in QuantEcon/workspace-lectures#20. AGENTS.md gains a one-line exception for the already-served case; the file must be resolved (permission, an open replacement, or removal) before data.quantecon.org is promoted as a public open-data host. The lecture only uses two of its columns (Country (en), Continent) for a continent lookup.

Follows

Repoints in the two lecture repos are separate PRs (data PR first, per the ordering rule); each will carry the one-time byte-compare gate in its body. Then P2 is ticked on the cutover tracker #15.

Part of QuantEcon/meta#338 (pilot) and QuantEcon/QuantEcon.manual#108 (convention).

[Copilot]

Pull request overview

Adds the “pandas_panel” P2 pilot datasets to the flat lectures/ namespace with provenance manifests, enabling two lecture series to consume a single canonical copy while recording integrity and licensing constraints.

Changes:

• Added realwage.csv, employ.csv, and countries.csv under lectures/, each with a sidecar manifest capturing provenance, integrity hashes, schema expectations, and consumers.
• Recorded inherited/no-builder status for the two constructed datasets via builder: null + builder_status: unrecovered.
• Updated AGENTS.md to document an explicit exception for caching already-publicly-served restricted datasets (with required manifest markings and tracking).

Reviewed changes

Copilot reviewed 5 out of 7 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
lectures/realwage.csv	Adds the OECD real minimum wages dataset snapshot used by the pandas panel lecture.
lectures/realwage.csv.yml	Documents provenance/licensing, integrity hash, schema, and dual consumers for realwage.csv.
lectures/employ.csv.yml	Documents provenance/licensing, integrity hash, schema, and dual consumers for employ.csv.
lectures/countries.csv	Adds the WorldData.info country reference table snapshot (semicolon-delimited) used for continent lookup.
lectures/countries.csv.yml	Documents restricted licensing status, integrity hash, schema, and dual consumers for countries.csv.
AGENTS.md	Documents the “already served publicly” restricted-file caching exception and required tracking.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.