Update PolicyEngine bundle to 6.3.4

[policyengine]

Fixes #3921
Fixes #3922

Refs #3864: state reports still use the national dataset under PolicyEngine/policyengine.py#604, so that issue stays open.

Summary

Update PolicyEngine .py bundle from 6.2.1 to 6.3.4.

Bundled versions

• policyengine: 6.3.4
• policyengine-core: 3.33.0
• policyengine-us: 2.2.1
• policyengine-uk: 2.126.12
• spm-calculator: 1.0.0

Certified data releases

• US: populace-us-2024-spm-20260915
• UK: policyengine-uk-data-1.58.0

Additional changes for 6.3.4

Added in 16e3c74, carried over from #3868, which this PR replaces:

• UK membership filters in the catalog (Catalog extraction fails on the UK England region's list-valued filter #3922). Catalog extraction failed on 6.3.4 with region 'country/england' has an unsupported filter value, because the England region's filter now lists all nine English regions. Flat lists are stored as a JSON array in the existing filter text column. Scalar filters keep their current form, and nested, empty, non-finite or over-255-character values are still refused. Unit tests cover each case, and a fixed changelog fragment is added.

• Installed-catalog expectations. The counts are updated for US 2.2.1 and UK 2.126.12: 7,294 variables, 29,300 parameter nodes, 104,158 parameters, 1,194,256 parameter values, 2 datasets and 826 regions. The test also checks the England filter. Following Temporarily filter US state and district regions from the national dataset policyengine.py#604, every US region defaults to the national dataset, and only places fall back to it.

• Stage 12 model test fixtures (07108a4, cherry-picked from 77b44b3 on Add a rules registry that makes every enforced and unenforced rule visible #3884). The fixtures fixed retention_expires_at at 14 October 2026 but left created_at on the real clock, so from that date every insert failed the retention_expires_at > created_at check and the Test job failed on every branch. The fixtures now set created_at and updated_at to the same fixed time. The change is identical on Add a rules registry that makes every enforced and unenforced rule visible #3884, so either PR can merge first.

Locally on the locked 6.3.4 install: mypy passes, the extraction unit tests pass (36), and the installed-catalog test passes with RUN_V2_CATALOG_COMPATIBILITY=1.

---

Generated automatically by GitHub Actions; the "Additional changes" section was added by hand.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 87.58%. Comparing base (83da680) to head (07108a4).
⚠️ Report is 10 commits behind head on master.

Additional details and impacted files

@@             Coverage Diff             @@
##           master    #3918       +/-   ##
===========================================
+ Coverage   45.22%   87.58%   +42.36%     
===========================================
  Files         183      198       +15     
  Lines       11073    12223     +1150     
  Branches     1905     2161      +256     
===========================================
+ Hits         5008    10706     +5698     
+ Misses       5644      921     -4723     
- Partials      421      596      +175     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.