Add embeddable dashcam recording-time calculator

[vortexopenclaw]

Current result

• 85 camera presets, with Brand → Model → Channel configuration and a recording-setting selector only when alternatives exist.
• Stacked rows on both desktop and mobile. Each card capacity shows ONE estimated recording time, rounded to five minutes.
• Prefer duration-weighted full-file sample rates where available; otherwise use midpoint bitrate. Preserve manufacturer duration tables as supplied. Underlying source ranges remain evidence, not displayed time ranges.
• Per-channel resolution/FPS and MB/min remain together. Exact-setup photos only; unmatched setups have no photo. No repeated site eyebrow, photo captions or affiliate disclosure inside the embed. No memory-card shopping links pending correct capacity-specific destinations.
• Static bundle, auto-resizing embed and WordPress shortcode ZIP. No website installation or deployment.

Recovered measurements and coverage

• Audited all 50 original submissions currently available, privately. Saved sanitized technical evidence: 264 paired driving-file measurements and 96 driving-summary groups, including repeated scans and interrupted clips. These are evidence counts, not unique-camera counts.
• Added Botslab G980H, Thinkware ARC 800, Thinkware FA200 and Ottocast OttoSafe Cam.
• Recovered padded full-loop rates for S1 QHD. Added a full-file default estimate for Cansonic Z4; named quality alternatives remain explicitly video-only where full-file pairs do not establish the menu selection.
• Constant file allocation can establish the size of an observed full-length loop. Unrelated size/duration extrema are never paired. No guessed padding multiplier.
• Five maintained-profile gaps remain: Escort MAXcam 360c, Rove R2-4K, TeslaCam, Thinkware ARC 700 and ARC 900. Broader catalog-only coverage is not complete. Unreleased H1 stays excluded.
• Exact evidence limits and mapping: web/recording-time/COVERAGE.md, coverage-2026-09-08.csv and storage-evidence.json. Canonical profile evidence is maintained in vortexopenclaw/dashcam-camera-profiles.

Source limits retained

• T800 regional 3CH manufacturer duration is retained without inventing rear-camera specifications or per-channel MB/min.
• Inconsistent A229 Plus Front-only Maximum manufacturer figures remain omitted.
• Unknown menu settings are not certified as maximum quality. Short preallocated clips are retained as evidence but excluded from full-loop continuous-recording estimates.

Verification and workflow

• 14 Python exporter tests and 6 Node calculation tests passed.
• PHP lint and shortcode contract passed in CI.
• Full 85-camera Chromium suite passed: configuration/setting changes, source attribution, single card-time estimates, four-channel labels, responsive layouts, photo selection/fallback, iframe sizing and missing-data behavior. Desktop/mobile previews delivered.
• One calculator CI run per PR update; main still gets post-merge verification. Public-content privacy check audits the actual PR head, not GitHub's synthetic merge author.
• Follow-up visual tweaks stay in this PR. App/scanner logging and release workflow corrections were merged separately in PR Preserve padding-inclusive storage measurements in card scans #20, without deploying or releasing them.

Publication state

GitHub source sharing is authorized. Raw submissions and credentials remain private. This PR remains a draft pending visual approval. No WordPress draft installation, website publication or app/receiver deployment has occurred. Website deployment is separate from building or merging code.