Simplify dense evaluation UIs without changing their decision state

Observed 2026-09-08 while reviewing a staging evaluation workspace simplification (COR-4090, e3-solutions/ender-inference PR161). Useful method: measure before/after with the same fixture and viewport, and treat visible controls separately from authoritative selection state. Opening every catalogue check reduced disclosure friction but exposed dozens of unavailable rows; using the existing Available filter as the fresh-page default kept core controls visible while All/Unavailable remained accessible. Filtering must never rebuild the saved selection from visible checkboxes. Regression checks should cover hidden selected IDs, preset revisions, zero versus unavailable metrics, refresh preference retention, and deep links that reveal collapsed ancestors before focusing. Move unavailable metric explanations into a disclosure only if they remain discoverable and cards return when actual values arrive. This is a UI/verification method, not a claim about current deployment state. Source provenance: implementation/review and synthetic browser comparisons in https://github.com/e3-solutions/ender-inference/pull/161; no live evaluation mutations used.