feat(ui): progress indicator while the cloud OCR reads a photo

chore(ui): drop GitHub link from footer (repo is private)

fix(ui): consent modal was trapped under the samples modal

feat(ui): preview sample documents before ingesting them

fix(graph): visible edges + measurement values in node labels

feat: escalation threshold derived from costs; /metrics exposes decision economics

fix(l2b): units + document date from photo extraction; drop date/id-shaped values

feat: help, document viewer, knowledge graph; empty start + language-matched demo

fix: show every extracted fact, its date, and the document's real language

Unit guard before L2a, LOINC aliases, cloud consent, honest confidence

Q&A: answer in English without the Ukrainian metric name

Tell the user the cloud OCR takes ~90 s instead of leaving them guessing

Photo OCR: fall back to image->facts when transcription yields none; surface empty extraction in UI

Per-session cards: each visitor gets an isolated seeded card (cookie lmc_sid)

Real OCR on image upload via cloud ZeroGPU Lab (text/seed stay deterministic)

Investor UX: full EN localization, story ribbon, signal grouping, provenance drawer

Sync with local: EN/UA UI toggle + provenance/span/retrieval

Deploy LiveMedCard demo (Docker, stub, seeded bilingual card)

initial commit