common-sign-language-0001: fix the task type

The task it performs is clearly more akin to action recognition than object
attribute estimation. In fact, its interface is almost exactly like that of
i3d-rgb-tf.

Author: Roman Donchenko

models/public/common-sign-language-0001/model.yml

@@ -17,7 +17,7 @@ description: >-
   person as an input and outputs logits of predicted classes. The model is based on
   the S3D-Mobilet-V3 backbone developed for fast inference and outputs an logits
   vector of 27 floats.
-task_type: object_attributes
+task_type: action_recognition
 files:
   - name: s3d-rgb-mobilenet-v3-large-stream-jester.onnx
     size: 16647750