[Jammy2211]

Model inspection now opens in a separate browser tab with Instrument, Device and Configuration selectors, Configuration details first, and compact timing bars. The dashboard shows curated float64 / 1500-source-pixel defaults where applicable; archived captures and direct links remain intact.

Removes the requested legacy controls and repetitive prose, standardizes model labels, and separates Profiling Results from Profiling scripts. Measurement sections directly select an available default when necessary, without joining unrelated recorded setups. Bars use a matching full likelihood or component total where recorded, otherwise an explicitly labeled common maximum; distinct units and measurement semantics retain separate scales.

Scripts Changed

• pulse/campaigns.py, pulse/board.py: landing-page cleanup and Results navigation.
• pulse/setup_browser.py, .js, .css: model detail routes, defaults, selectors, labels and charts.
• dashboard.html, dashboard.md: regenerated from unchanged committed captures.
• Browser and renderer tests cover navigation, archived links, precision/source-size filtering, measurement scales and immutable evidence loading.

Validation

• 193 Python tests passed.
• Chromium: new-tab navigation, reloads, selectors, archived links, hash mismatch/retry, inline evidence, scale separation and proportional bars.
• Responsive checks at 390, 768, 820, 1024 and 1440 px in light/dark.
• Full captured Delaunay/A100 preview verified against exact-commit producer shards.
• Ruff lint and format, git diff --check, and offline catalogue/dashboard/feed checks passed.
• No profiling runs submitted; no raw result/capture changes.

Readiness

User authorized this development ship: “I authorize, maybe we should get those repos up to date first?” All four libraries were verified clean on main and exactly at fetched origin/main; no pull was needed. Heart was refreshed and all prior RED reasons cleared before shipping. No RED override was exercised.

Current Heart verdict: YELLOW.

• manifest drift: workspace checkouts (manifest ↔ disk) — 2 mismatch(es) vs PyAutoMind/repos.yaml
• Stale evidence: release validation incomplete: no rehearsal for current source

This is development shipping only; merge remains a separate human action.

Tier: judge — merge mode: human /prm.

Closes #25