test(spin): the native-spin graph .pt2 must carry the with-comm artifact

test_native_spin_graph_freeze asserted has_comm_artifact is True and, six
lines later, that forward_lower_with_comm.pt2 is ABSENT. Those contradict
each other. The metadata assertion was updated when native spin gained
multi-rank support on the graph lower (804cc57); the file assertion is a
leftover from the single-rank contract that preceded it.

Caught only on GPU: the CPU freeze fails earlier on this workstation's
inductor AVX2 codegen bug, so the stale assertion was never reached
locally, and Test CUDA runs only on the merge queue.

Invert it -- the nested artifact must be present -- so the two checks agree
and a regression that silently stopped embedding it would fail here.

Author: Han Wang

source/tests/pt_expt/model/test_dpa4_export.py

@@ -466,7 +466,13 @@ def test_native_spin_graph_freeze(tmp_path) -> None:
     assert "force_mag" in md["output_keys"]
     for key in ("atom_energy", "energy", "force", "virial"):
         assert key in md["output_keys"]
-    assert not any(n.endswith("forward_lower_with_comm.pt2") for n in names)
+    # ... and the nested artifact must actually BE there. This assertion was
+    # inverted while native spin was single-rank; leaving it that way made the
+    # test contradict its own has_comm_artifact check above, and it survived
+    # only because the CPU freeze fails earlier on an unrelated inductor bug.
+    assert any(n.endswith("forward_lower_with_comm.pt2") for n in names), (
+        "has_comm_artifact is True but the nested with-comm .pt2 is missing"
+    )
 
 
 def test_native_spin_nlist_deserialize_rejected(tmp_path) -> None: