ci: add manual OCR integration workflow

.github/workflows/ocr-integration.yml

@@ -0,0 +1,34 @@
+name: OCR integration
+
+on:
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  ocr-integration:
+    name: Real PaddleOCR integration
+    runs-on: ubuntu-latest
+    timeout-minutes: 20
+    defaults:
+      run:
+        working-directory: services/ocr-service
+    env:
+      DEBUG: 'false'
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+          cache: pip
+          cache-dependency-path: services/ocr-service/requirements.txt
+
+      - name: Install dependencies
+        run: python -m pip install --upgrade pip && python -m pip install -r requirements.txt
+
+      - name: Run real OCR integration tests
+        run: python -m pytest -m integration --runintegration --runslow -v

services/ocr-service/README.md

@@ -252,6 +252,19 @@ pytest --cov=app --cov-report=html
 pytest tests/test_engine.py -v
 ```
 
+### Real OCR Integration
+
+The deterministic suite intentionally does not download PaddleOCR models. The
+real engine test is isolated in the `OCR integration` workflow and can be
+started manually from GitHub Actions. Locally, run:
+
+```bash
+pytest -m integration --runintegration --runslow -v
+```
+
+This test is slower and requires model downloads; it is not a required pull
+request check.
+
 ### Manual Testing
 
 ```bash