Save TorchScripted model for LinearBandit

Summary: Change Pearl FBLearner Flow workflow to save TorchScripted model for `LinearBandit`.

Reviewed By: PavlosApo

Differential Revision: D65865605

fbshipit-source-id: a33fceeaf85382c3880318a2b6342491b3180399

pearl/replay_buffers/tensor_based_replay_buffer.py

@@ -262,7 +262,7 @@ def sample(self, batch_size: int) -> TransitionBatch:
         """
         if batch_size > len(self):
             raise ValueError(
-                f"Can't get a batch of size {batch_size} from a replay buffer with"
+                f"Can't get a batch of size {batch_size} from a replay buffer with "
                 f"only {len(self)} elements"
             )
         samples = random.sample(self.memory, batch_size)