fix FastSAC evaluate_policy missing set_is_evaluating call

Author: maedmatt

src/holosoma/holosoma/agents/fast_sac/fast_sac_agent.py

@@ -1000,6 +1000,7 @@ def export(self, onnx_file_path: str) -> None:
 
     @torch.no_grad()
     def evaluate_policy(self, max_eval_steps: int | None = None):
+        self.env.set_is_evaluating()
         obs = self.env.reset()
 
         for _ in itertools.islice(itertools.count(), max_eval_steps):