Delete the migrated test_pwt.csv copy (Track D settle)

[mmcky]

Step 2 of the settle-week plan (QuantEcon/workspace-lectures#48), gated until now on the four republishes: en and zh-cn published publish-2026aug19 earlier today, and fr/fa followed once the jax 0.11.1 regression was pinned away (QuantEcon/workspace-lectures#49). All four sites were verified on 2026-08-19 to carry only the data-lectures URL in their rendered pandas/polars output — en and zh-cn additionally in their published _notebooks — with zero references to the path this PR deletes (fr/fa publish no downloadable notebooks, so their rendered HTML is the whole verification surface).

Deletion-time org sweep (run 2026-08-19): Trees-API basename pass over 278/278 QuantEcon org repos, 0 fetch failures, positive and negative controls fired; plus a content pass over every gh search code candidate with the two private repos fetched authenticated. No repo reads this URL. The only URL-form readers found anywhere (africa/workshop course repos, indian workshop) point at the pre-myst master/source/… layout, which already serves 404 — pre-existing decay in the class recorded by QuantEcon/workspace-lectures#38, unaffected by this deletion. The .ml translation's local copy is read by nothing and stays for the Track X orphan sweep, per decision 2 on QuantEcon/workspace-lectures#48.

Preview note: this PR's preview is the prune-discrimination capture for QuantEcon/workspace-lectures#41 — the restored cache was built Monday 08-17 and physically contains this file, so with the #616 prune the preview should 404 it while serving controls. Probe results will be posted in this thread while the deploy is alive.

After merge: the four live sites keep serving their baked _static copies until each one's next publish — that clearance belongs to the ongoing scope of QuantEcon/workspace-lectures#40, not this PR. The canonical file remains at data-lectures (lectures/test_pwt.csv, sha256-manifested).

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Removes the legacy/migrated test_pwt.csv asset from this lecture repo’s _static tree, aligning with the migration plan to source the dataset exclusively from QuantEcon/data-lectures.

Changes:

• Deleted lectures/_static/lecture_specific/pandas/data/test_pwt.csv.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[github-actions]

🚀 Deployed on https://6a85267fe62fc7649f51a8d0--epic-agnesi-957267.netlify.app

[mmcky]

Preview discrimination evidence for the QuantEcon/workspace-lectures#41 rollout, captured 2026-08-19 while the deploy is alive (deploy 6a85267f…, run 32212747847, head 55a0b53).

The restored cache provably contains the deleted file: this preview restored the newest build-cache artifact (built Monday 2026-08-17, before this deletion), and the live site — deployed from that same artifact lineage — still serves the file right now (python-programming.quantecon.org/…/test_pwt.csv → 200, 793 B). Stale bytes in, clean site out:

probe (deploy-id URL form)	result
/_static/lecture_specific/pandas/data/test_pwt.csv (deleted here)	404 (3,449 B Netlify 404 page)
/pandas.html (control)	200, 189,964 B
/_static/lecture_specific/troubleshooting/launch.png (control, restored from the same cache)	200, 1,846 B

That is the #616 prune working in isolation: without it, the restored _build/html would have republished the deleted csv into this preview.

URL-form check, re-derived live per the method note on QuantEcon/workspace-lectures#41: the deploy-preview-618-- alias form 404s everything on this site too — including the pandas.html control (404, same 3,449 B body as the deleted-file probe) — while the deploy-id-prefixed form serves normally. So a probe against the alias form would read "deletion propagated" when it means "wrong URL"; the finding from the first rollout repo generalizes to this site.