Improve calibration target pages and investigation guidance

[anth-volk]

Fixes #161

Summary

• streamline individual calibration-target cards into one Target details section organized around definition, scope, and official source
• remove redundant generated calculations, model-representation details, internal identifiers, lineage fields, and the release-artifacts panel
• add a release-aware View on Hugging Face action beside the dataset release selector
• present targets without generated model constraints as Skipped in the Calibration Fit chart
• treat targets without geography metadata as United States / National and remove the missing-geography presentation and query path
• move calibration-target investigation guidance into tool-independent documentation referenced by both AGENTS.md and CLAUDE.md
• document the required local Cross-dataset artifact configuration

Why

The calibration pages had accumulated implementation and lineage fields that obscured the target definition and source information most useful to readers. They also exposed producer-specific distinctions and missing metadata directly: targets that produce the same unavailable calibration result appeared under separate statuses, while nationally scoped internal targets without explicit geography fields appeared unclassified.

This change simplifies the presentation, normalizes those dashboard concepts, and preserves detailed investigation procedures in shared documentation instead of tying them to one AI coding tool.

Validation

• bun test — 188 tests passed
• bun run lint — TypeScript validation passed

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
calibration-diagnostics	Ready	Preview	Aug 19, 2026 7:12pm