refactor: add support to outputs key in inference request inputs

ts/torch_handler/request_envelope/kservev2.py

@@ -113,6 +113,14 @@ def _from_json(self, body_list):
                 )
             input_names.append(input["name"])
         setattr(self.context, "input_names", input_names)
+
+        output_names = []
+        for index, output in enumerate(body_list[0].get("outputs", [])):
+            output_names.append(output["name"])
+            # TODO: Add parameters support
+            # parameters = output.get("parameters")
+        setattr(self.context, "output_names", output_names)
+
         logger.debug("Bytes array is %s", body_list)
         id = body_list[0].get("id")
         if id and id.strip():
@@ -167,10 +175,15 @@ def _batch_to_json(self, data):
         Splits batch output to json objects
         """
         output = []
-        input_names = getattr(self.context, "input_names")
+
+        output_names = getattr(self.context, "output_names")
+        delattr(self.context, "output_names")
+        if len(output_names) == 0:
+            # Re-use input names in case no output is specified
+            output_names = getattr(self.context, "input_names")
         delattr(self.context, "input_names")
         for index, item in enumerate(data):
-            output.append(self._to_json(item, input_names[index]))
+            output.append(self._to_json(item, output_names[index]))
         return output
 
     def _to_json(self, data, input_name):