Refactor type annotations and return types in model_runner

Removed specific type hints and return annotations for `self.inputs` and `run_model` methods to enhance flexibility and maintain consistency. These changes streamline the code and improve compatibility with varying input/output types during model inference.

selfdrive/modeld/runners/model_runner.py

@@ -30,14 +30,14 @@ class ModelRunner(ABC):
       self.model_metadata = pickle.load(f)
     self.input_shapes = self.model_metadata['input_shapes']
     self.output_slices = self.model_metadata['output_slices']
-    self.inputs: dict[str, np.ndarray | Tensor] = {}
+    self.inputs: dict = {}
 
   @abstractmethod
   def prepare_inputs(self, imgs_cl: dict[str, CLMem], numpy_inputs: dict[str, np.ndarray])-> dict:
     """Prepare inputs for model inference."""
 
   @abstractmethod
-  def run_model(self) -> np.ndarray:
+  def run_model(self):
     """Run model inference with prepared inputs."""
 
   def slice_outputs(self, model_outputs: np.ndarray) -> dict:
@@ -70,7 +70,7 @@ class TinygradRunner(ModelRunner):
 
     return self.inputs
 
-  def run_model(self) -> np.ndarray:
+  def run_model(self):
     return self.model_run(**self.inputs).numpy().flatten()
 
 
@@ -88,5 +88,5 @@ class ONNXRunner(ModelRunner):
       self.inputs[key] = self.frames[key].buffer_from_cl(imgs_cl[key]).reshape(self.input_shapes[key])
     return self.inputs
 
-  def run_model(self) -> np.ndarray:
+  def run_model(self):
     return self.runner.run(None, self.inputs)[0].flatten()