force the seed in some broken tests

lightsim2grid/tests/test_dist_slack_backend.py

@@ -91,14 +91,20 @@ def test_after_reset(self):
         _ = self.env_ds.reset()
         self._prepare_env(self.env_ds)
         self._aux_test_different(self.env_ss, self.env_ds)
-
+
+    def _aux_get_kwargs_runner(self):
+        return dict(nb_episode=1, max_iter=self.max_iter_real, add_detailed_output=True, seed=0)
+
     def test_after_runner(self):
         """test I can use the runner"""
         runner_ss = Runner(**self.env_ss.get_params_for_runner())
         runner_ds = Runner(**self.env_ds.get_params_for_runner())
-        res_ss = runner_ss.run(nb_episode=1, max_iter=self.max_iter_real, add_detailed_output=True)
-        res_ds = runner_ds.run(nb_episode=1, max_iter=self.max_iter_real, add_detailed_output=True)
-        assert res_ss[0][3] == res_ds[0][3], f"{res_ss[0][3]} vs {res_ds[0][3]}"  # same number of steps survived
+        res_ss = runner_ss.run(self._aux_get_kwargs_runner())
+        res_ds = runner_ds.run(self._aux_get_kwargs_runner())
+        import pdb
+        pdb.set_trace()
+        if res_ss[0][3] != res_ds[0][3]: # same number of steps survived
+            raise RuntimeError(f"{res_ss[0][3]} vs {res_ds[0][3]}: ")
         assert res_ss[0][2] != res_ds[0][2]  # not the same reward
         ep_ss = res_ss[0][-1]
         ep_ds = res_ds[0][-1]