Add SNAP congressional district validation diagnostics

Problem

Populace now supports congressional-district geography and CD calibration targets from the unified US file, but the current SNAP target surface is only USDA national/state benefit spending. ACS S2201 congressional-district SNAP household estimates are already registered in source coverage as snap_local_proxy / census-acs-s2201-congressional-district-snap-2024, but they are validation-only and are not surfaced as a release diagnostic.

That makes it easy for consumers to assume that SNAP congressional-district breakdowns are program-specifically calibrated when they are not.

Proposed work

Add a validation-only SNAP local diagnostics artifact that compares Populace district-level SNAP outputs/support with ACS S2201 district SNAP household estimates.

Suggested fields by congressional district:

• congressional district GEOID and display label
• ACS S2201 SNAP household estimate, and margin of error if available
• Populace weighted SNAP-recipient households by CD
• optional formula SNAP dollars by CD, if cheap enough at build time
• raw positive SNAP sample household count
• positive-recipient ESS / weight concentration diagnostics
• repeated state SNAP spending fit/error for context
• validation flags such as low support, outside ACS MOE, or state outlier

Architecture note

This should remain validation-only initially. It should not promote ACS S2201 to a hard calibration target until we understand measurement differences and interactions with USDA state benefit spending targets.

This matches the existing source coverage architecture, where validation-only families may appear in diagnostics but should fail if activated as hard targets.

Motivation

A recent SNAP audit found that the current release has congressional-district geography and CD targets, but no SNAP-specific CD target fit. State SNAP spending has a meaningful error tail, and raw SNAP-positive support is thin in many districts, so district SNAP outputs need explicit diagnostics before being used for detailed CD analysis.

Start by inspecting the existing source coverage architecture, the snap_local_proxy registration, and how validation-only families are surfaced in release diagnostics. Define the district-level artifact around the listed ACS S2201 and Populace measures, with validation flags and state-fit context, while ensuring it cannot become a hard calibration target until measurement differences are understood.