Add results.json schema validation and source tracking

[PavelMakarchuk]

Summary

Adds a policyengine.results module with two small pieces that support the blog post content pipeline:

• Schema validation (schema.py): Pydantic models that validate results.json at generation time
• Source tracking (tracking.py): Helper that auto-captures line numbers for traceability

What it does

Schema validation

from policyengine.results import ResultsJson, ResultsMetadata, ValueEntry

results = ResultsJson(
    metadata=ResultsMetadata(title="SALT Cap Repeal", repo="PolicyEngine/analyses"),
    values={
        "budget_impact": ValueEntry(
            value=-15.2e9,
            display="$15.2 billion",
            source_line=47,
            source_url="https://github.com/.../analysis.py#L47",
        ),
    },
)
results.write("results.json")

Catches errors at generation time:

• Missing source_line or source_url on any value
• Table rows with wrong column count
• Chart alt text that's too short (< 20 chars)
• Missing required metadata fields

Source tracking

from policyengine.results import tracked_value

budget = reform_revenue - baseline_revenue
results["values"]["budget_impact"] = tracked_value(
    value=budget,
    display=f"${abs(budget)/1e9:.1f} billion",
    repo="PolicyEngine/analyses",
)
# Automatically captures source_line and builds source_url

Why

Every number in a PolicyEngine blog post should link to the exact line of code that produced it. This module ensures the results.json contract is valid before it reaches the resolve-posts build step.

Test plan

• 11 unit tests passing (schema validation, source tracking, edge cases)
• CI passes

🤖 Generated with Claude Code