Allow use of prior cached model

Moving the requests.get(url) below the check for cache.  If your vision system is offline, and you have a cached model, the check prior to checking cache throws an exception.  Now if you have a cached model it will use that first and only if you dont have a model will it switch to checking api.roboflow.com

depthai_sdk/src/depthai_sdk/integrations/roboflow.py

@@ -27,16 +27,18 @@ def _file_with_ext(self, folder: Path, ext: str) -> Optional[Path]:
 
     def device_update(self, device: dai.Device) -> Path:
         mxid = device.getMxId()
-        url = f"https://api.roboflow.com/depthai/{self.config['model']}/?api_key={self.config['key']}&device={mxid}"
-        response = requests.get(url)
-
         name = self.config['model'].replace('/', '_')  # '/' isn't valid folder name
 
         model_folder = ROBOFLOW_MODELS / name
         json_file = self._file_with_ext(model_folder, '.json')
         if json_file:
             return json_file
 
+        #Check the URL after checking the cache to make sure vision systems that are offline don't throw an exception
+        url = f"https://api.roboflow.com/depthai/{self.config['model']}/?api_key={self.config['key']}&device={mxid}"
+        response = requests.get(url)
+
+
         json_res = response.json()
         if "error" in json_res:
             raise Exception(json_res['error'])