Bundle the whisper base model + one-shot standalone provisioning

- models/hf-cache/ vendors Systran/faster-whisper-base (142MB, gitignored);
  engine spawn pins HF_HUB_CACHE there, so transcription works with a cold
  ~/.cache and fully offline
- verified: HOME=/tmp/fakehome HF_HUB_OFFLINE=1 model loads and transcribes
  via the bundled cache alone
- setup-standalone.sh: idempotent fresh-clone provisioning (local venv,
  model copy-or-download, engine boot smoke on :8123) — runs green
- backend/README rewritten for the desktop bundled-lite reality (was
  pointing at a nonexistent root README/docs and a Postgres quickstart)

Kotlin suite still green (gradlew test, 0 failures)

setup-standalone.sh

@@ -0,0 +1,61 @@
+#!/usr/bin/env bash
+# One-shot provisioning for a fresh clone of SHONAR Desktop:
+# local engine venv + vendored whisper base model. Idempotent.
+set -uo pipefail
+cd "$(dirname "$0")"
+
+echo "==> backend venv (local, editable install of THIS backend)"
+if [ ! -x backend/.venv/bin/uvicorn ]; then
+  command -v uv >/dev/null || { echo "uv not found — install it: curl -LsSf https://astral.sh/uv/install.sh | sh"; exit 1; }
+  (cd backend && uv venv .venv --python 3.11 && uv pip install -p .venv/bin/python -e ".[dev,faster-whisper]")
+else
+  echo "    already present"
+fi
+
+echo "==> whisper base model -> models/hf-cache"
+MC="models/hf-cache/models--Systran--faster-whisper-base"
+if [ -d "$MC/snapshots" ] && [ -n "$(ls -A "$MC/snapshots" 2>/dev/null)" ]; then
+  echo "    already present"
+else
+  # Prefer copying an existing local copy (works offline)...
+  src=""
+  for c in "${HF_HUB_CACHE:-$HOME/.cache/huggingface/hub}/models--Systran--faster-whisper-base" \
+           "$HOME/.cache/huggingface/hub/models--Systran--faster-whisper-base"; do
+    [ -d "$c/snapshots" ] && { src="$c"; break; }
+  done
+  mkdir -p models/hf-cache
+  if [ -n "$src" ]; then
+    echo "    copying from $src"
+    cp -r "$src" models/hf-cache/
+  else
+    echo "    no local copy — downloading from HuggingFace (needs internet once)"
+    backend/.venv/bin/python - <<'EOF'
+from huggingface_hub import snapshot_download
+snapshot_download("Systran/faster-whisper-base", cache_dir="models/hf-cache")
+EOF
+  fi
+fi
+
+echo "==> smoke: engine boots on :8123 with the bundled cache"
+tmp=$(mktemp -d)
+SHONAR_SECRET_KEY="bootstrap-verify-secret-0123456789abcdef" \
+SHONAR_DATABASE_URL="sqlite+aiosqlite:///$tmp/e.db" \
+SHONAR_QUEUE_BACKEND=inline SHONAR_AUTO_MIGRATE=1 \
+SHONAR_STORAGE_PATH="$tmp/storage" \
+HF_HUB_CACHE="$PWD/models/hf-cache" \
+backend/.venv/bin/uvicorn shonar.main:app --app-dir backend --port 8123 \
+  >"$tmp/uvicorn.log" 2>&1 &
+pid=$!
+ok=""
+for _ in $(seq 1 30); do
+  sleep 1
+  curl -sf -m 2 http://127.0.0.1:8123/api/v1/readyz >/dev/null && { ok=1; break; }
+done
+kill $pid 2>/dev/null; wait $pid 2>/dev/null
+rm -rf "$tmp"
+if [ -n "$ok" ]; then
+  echo "    engine ready — provisioning complete. Start the app: ./desktop-dev.sh"
+else
+  echo "    ENGINE FAILED TO START" >&2
+  exit 1
+fi