Deprecate split state/district H5 pipeline after Populace replacement

[MaxGhenis]

Populace is moving to a single national H5 with state and congressional-district region scoping in policyengine.py rather than one-file-per-state and one-file-per-district artifacts. Once the new Populace release is certified into policyengine.py, this repo should stop presenting states/*.h5 and districts/*.h5 as the live path.

Tracking context:

• Replace all remaining policyengine-us-data artifacts with Populace microcosm#204 is the central replacement tracker.
• Publish US releases as a single national H5 microcosm#237 removed state/district artifact generation from Populace releases.
• Current downstream split-H5 consumers are listed in Replace all remaining policyengine-us-data artifacts with Populace microcosm#204 comment: Replace all remaining policyengine-us-data artifacts with Populace microcosm#204 (comment)

Cleanup checklist:

• Mark state/district H5 production as legacy or archived after Populace certification.
• Stop publishing new split regional H5 files unless explicitly needed for historical reproducibility.
• Keep enough manifests/docs for old releases to remain reproducible.
• Redirect docs and examples to the certified policyengine.py Populace bundle and region/scoping APIs.
• Remove or quarantine release-promotion tests that assume states/ and districts/ are active release outputs.

[MaxGhenis]

Pre-archive sweep complete — 2026-07-02

Closed all open issues and PRs ahead of archiving this repository, per the acceptance criteria in PolicyEngine/microcosm#204 (each item "linked to Populace issues, explicitly marked benchmark-only/out-of-scope, or closed as obsolete").

Issues: 146 closed as not planned, 0 closed as completed. (No genuinely-shipped-but-unclosed items were found on review.) All closed with a standard comment linking to PolicyEngine/microcosm#204; 119 of the 146 also link a specific Populace/Ledger counterpart (drawn primarily from PolicyEngine/microcosm#46, Max's own 2026-06-14/06-27 triage audit, which enumerates destinations for most of this backlog). The remaining 27 got the #204-only link — mostly dead old-repo infra (Stage 5 diagnostics, HDFStore format, ReactFlow diagrams, GCS smoketest) or enhancement requests already subsumed by Populace's target/calibration architecture.

PRs: 27 closed, 0 merged. None were approved or CI-green; several had merge conflicts. Per populace#46: "old staged-pipeline service-boundary PRs... should not be merged as a path to replacement." PR #1177 (archive banner) was merged separately in this same window — not touched by this sweep.

Left open: only this issue (#1175).

Genuinely unresolved data-quality items with no populace counterpart found: none. Every substantive open bug checked (#349 household_id aggregation, #444 H5 caching, #555 19x inflated state aggregates, #557 Medicaid enrollment, #590 QRF batching, #1105/#1107 CBO overshoot, #1139/#1158/#1160 PUF clone weight) already has a live Populace tracker per populace#46/#47/#48/#49/#204 — no gaps requiring new populace issues.

Full issue-by-issue counterpart mapping available on request.

[MaxGhenis]

Archiving the repository now: README banner merged (#1177), all open issues (146) and PRs (27) closed with pointers per PolicyEngine/microcosm#204's criteria. Published PyPI/HF/GCS artifacts remain available, frozen. Historical work continues in PolicyEngine/populace.