[AZ-180] Refactor inference and engine factory for improved model handling

- Updated the autopilot state to reflect the current task status as in progress.
- Refactored the inference module to streamline model downloading and conversion processes, replacing the download_model method with a more efficient load_source method.
- Introduced asynchronous model building in the inference module to enhance performance during model conversion.
- Enhanced the engine factory to include a new method for building and caching models, improving error handling and logging during the upload process.
- Added calibration cache handling in the Jetson TensorRT engine for better resource management.

Made-with: Cursor

src/engines/jetson_tensorrt_engine.pyx

@@ -1,5 +1,39 @@
+import os
+import tempfile
 from engines.tensorrt_engine cimport TensorRTEngine
+from loader_http_client cimport LoaderHttpClient, LoadResult
 
 
 cdef class JetsonTensorRTEngine(TensorRTEngine):
-    pass
+    @staticmethod
+    def convert_from_source(bytes onnx_model, LoaderHttpClient loader_client, str models_dir):
+        cdef str calib_cache_path
+        calib_cache_path = JetsonTensorRTEngine._download_calib_cache(loader_client, models_dir)
+        try:
+            return TensorRTEngine.convert_from_source(onnx_model, calib_cache_path)
+        finally:
+            if calib_cache_path is not None:
+                try:
+                    os.unlink(calib_cache_path)
+                except Exception:
+                    pass
+
+    @staticmethod
+    def _download_calib_cache(LoaderHttpClient loader_client, str models_dir):
+        cdef LoadResult res
+        import constants_inf
+        try:
+            res = loader_client.load_big_small_resource(
+                constants_inf.INT8_CALIB_CACHE_FILE, models_dir
+            )
+            if res.err is not None:
+                constants_inf.log(f"INT8 calibration cache not available: {res.err}")
+                return None
+            fd, path = tempfile.mkstemp(suffix=".cache")
+            with os.fdopen(fd, "wb") as f:
+                f.write(res.data)
+            constants_inf.log("INT8 calibration cache downloaded")
+            return path
+        except Exception as e:
+            constants_inf.log(f"INT8 calibration cache download failed: {str(e)}")
+            return None