Feature/run jetson e2e tests (#4)

* Run tests

* Run tests

* Run tests

* Run tests

* Added rebuild

* Added files for e2e tests

* Added rebuild

* Added rebuild

* Added biuld TensorRT flag

* Changed to use NumPy 1.x for Jetson

* Make universal invocation

* Make Cython constans

* Changed to prepare onnx

* Changed smoke-test to wait AI conversion

* Added step for model conversion

* Changed to not run step in parallel

* Push model to docker registry

* Push model to docker registry

* Push model to docker registry

.woodpecker/e2e-convert-jetson.yml

@@ -0,0 +1,30 @@
+when:
+  - event: [manual]
+    evaluate: 'E2E_CONVERT_JETSON == "1"'
+
+labels:
+  platform: arm64
+
+steps:
+  - name: e2e-convert-jetson
+    image: docker
+    environment:
+      REGISTRY_HOST:
+        from_secret: registry_host
+      REGISTRY_USER:
+        from_secret: registry_user
+      REGISTRY_TOKEN:
+        from_secret: registry_token
+    commands:
+      - apk add --no-cache bash docker-cli-compose
+      - echo "$REGISTRY_TOKEN" | docker login "$REGISTRY_HOST" -u "$REGISTRY_USER" --password-stdin
+      - cd e2e
+      - >
+        E2E_PROFILE=jetson
+        E2E_WAIT_FOR_ENGINE_ENABLED=1
+        E2E_ENGINE_WAIT_TIMEOUT=3600
+        E2E_LOG_TAIL=300
+        bash run_test.sh tests/test_health_engine.py::TestHealthEngineStep03Warmed
+      - bash scripts/publish_jetson_engine.sh
+    volumes:
+      - /var/run/docker.sock:/var/run/docker.sock