Fixed bug in dataset.py

- unsqueeze does not exist in NumPy
- now using the array backend

Author: boris-il-forte

mushroom_rl/core/dataset.py

@@ -477,7 +477,7 @@ def compute_J(self, gamma=1.):
         r_ep = split_episodes(self.last, self.reward)
 
         if len(r_ep.shape) == 1:
-            r_ep = r_ep.unsqueeze(0)
+            r_ep = self._array_backend.expand_dims(r_ep, 0)
         if self._dataset_info.backend == 'torch':
             js = self._array_backend.zeros(r_ep.shape[0], dtype=r_ep.dtype, device=r_ep.device)
         else: