fix: pass mae parameter to Paddle eval_pd in consistency tests

The eval_pd methods in TestEner and TestEnerGF were missing the mae
parameter, causing key mismatch when comparing Paddle output against
the reference (dpmodel) which includes mae_* keys.

Author: Han Wang

source/tests/consistent/loss/test_ener.py

@@ -297,6 +297,7 @@ def eval_pd(self, pd_obj: Any) -> Any:
             label,
             self.natoms,
             self.learning_rate,
+            mae=self.mae,
         )
         loss = to_numpy_array(loss)
         more_loss = {kk: to_numpy_array(vv) for kk, vv in more_loss.items()}
@@ -531,6 +532,7 @@ def eval_pd(self, pd_obj: Any) -> Any:
             label,
             self.natoms,
             self.learning_rate,
+            mae=True,
         )
         loss = to_numpy_array(loss)
         more_loss = {kk: to_numpy_array(vv) for kk, vv in more_loss.items()}