Add coverage-universe targets for uninsured and private coverage

[daphnehanse11]

Context

The healthcare tab currently focuses on ACA, Medicaid/CHIP, and Medicare program targets. It does not yet include the broader health coverage universe that explains what people have when they are not in those programs.

Missing coverage-universe targets include uninsured, employer-sponsored insurance, direct-purchase/private coverage, and related mutually-exclusive or overlapping coverage categories.

Goal

Add coverage-universe targets/diagnostics so healthcare release comparisons can distinguish program-specific misses from overall health insurance coverage composition misses.

Suggested implementation

• Inventory available source support, likely from CPS/ACS/Census or other health insurance coverage tables already flowing through Ledger/Populace.
• Decide whether targets should be person-level annual coverage, interview/current coverage, or both, matching PolicyEngine-US variable semantics.
• Add target roles for feasible concepts, such as:
  • uninsured
  • employer-sponsored insurance
  • direct-purchase/private coverage
  • Marketplace vs non-Marketplace direct purchase where source support allows
• Add diagnostics for category overlap/mutual-exclusivity so totals are interpretable.

Acceptance criteria

• Release artifacts expose coverage-universe target availability and source semantics.
• Tests cover mapping to the intended PolicyEngine-US variables or explicitly document formula/source ownership boundaries.
• Healthcare diagnostics can show whether program-target errors are accompanied by broad coverage-composition errors.
• Missing categories are recorded as unavailable or deferred rather than silently absent.

[daphnehanse11]

Relevant to the open concept decision here ("person-level annual coverage, interview/current coverage, or both"): Urban's HIPSM deliberately uses point-in-time/current coverage throughout — survey anchor, CMS admin snapshot targets, and model concept all at the same instant — and explicitly rejected annual averages for Medicaid targets (HIPSM 2020 methodology). #332 proposes making that convention contract doctrine for enrollment take-up flags, and #331 scopes the Medicaid/CHIP take-up materialization that would consume these coverage-universe targets. Recommend coverage-universe targets adopt the same current-coverage concept where source support allows, so program targets (month-tagged cms_medicaid snapshots, per #321) and universe targets stay comparable.