CD record sets stamp 22incd.csv as ty2023 — a tax-year vintage error; add Table 1.4 taxable-interest facts (populace#489)

Filed from the populace#489 target-truth adjudication (full receipts; populace-side fix in populace#518). Three corrections, all with official sourcing:

1. The congressional-district record sets mis-stamp the tax year (vintage error)

The v9.2 consumer feed (61b115c0…) carries irs_soi.ty2023.congressional_district_2022.* facts extracted from 22incd.csv with period: {tax_year: 2023}. That file is tax year 2022 data:

• IRS's CD data page lists tax year 2022 as the most recent congressional-district publication — there is no TY2023 CD release. "Congressional districts of 2022" is the boundary vintage (118th Congress), not tax-year+1.
• Arithmetic check against Pub 1304: the file's US taxable interest ($123.791B, A00300) is 92.7% of the Table 1.4 TY2022 actual ($133.597B, 22in14ar.xls) — the normal small-area coverage shortfall — but 39.4% of the TY2023 actual ($313.813B, 23in14ar.xls), impossible as a TY2023 all-returns tabulation. Return counts agree: 48.55M vs TY2022's 49.74M (−2.4%) vs TY2023's 55.26M (−12.1%).

Consequence in populace: the mis-stamp made the CD US aggregate look like the latest national interest fact, so the certified Build N surface rebased 457 state-AGI interest bins onto a TY2022 level it believed was TY2023, then under-aged it by one CBO link. populace#518 stops using CD aggregates as interest controls regardless, but the stamp should be corrected at source — every measure in these record sets is affected, not just interest (dollar measures on other concepts inherit the same one-link aging error wherever they serve as controls or targets).

2. Add Pub 1304 Table 1.4 taxable-interest facts (TY2022 + TY2023)

The feed has no Table 1.4 interest facts — the only Pub 1304 interest total is Table 4.3 (all_returns_excluding_dependents, $313.120B / 54,167,410 returns TY2023, 23in43ts.xls). populace#518 bridges that universe with a declared metadata stamp (dependent sliver: 0.22% of amount / 2.0% of returns at TY2023). Adding the Table 1.4 rows makes the bridge exact and retires it automatically (populace's control lattice prefers the exact all_returns universe):

• TY2022: taxable interest $133,596,569k / 49,736,855 returns (22in14ar.xls, "All returns, total", cols 19–20)
• TY2023: taxable interest $313,812,674k / 55,260,238 returns (23in14ar.xls)

These also give the aging model's chained-factor policy an interest actuals segment (_CBO_INCOME_SOURCE_TO_SOI_CHAIN in populace's target_aging.py — interest currently falls back to the CBO AGI default, which applied ×1.1203 over 2022→2024 while the realized TY2022→TY2023 growth was ×2.349; that miss is the root cause of populace#489).

3. Ingest HT2 TY2023 when IRS publishes it

HT2 is a correct photo of its own year (US A00300 vs Table 1.4: −0.36% TY2022, +0.13% TY2021 — verified against 22in55cmcsv.csv / 21in55cmcsv.csv). A TY2023 HT2 restores state-level interest truth, which no current fact provides — until then populace preserves TY2022 state shares under the national realized factor.

Start by locating the congressional-district record-set mappings and Pub 1304 ingestion entry points in chronicle, then read populace's target_aging.py where interest source chaining is referenced. Correct the CD tax-year metadata, add the TY2022 and TY2023 Table 1.4 taxable-interest facts, and define the HT2 TY2023 ingestion path for when the IRS publishes it; verify the official source values and affected measures.