BUG: testing: convert torch tensors with conjugate bit set to NumPy (#744)

Signed-off-by: Pradyot Ranjan <99216956+pradyotRanjan@users.noreply.github.com>
Co-authored-by: Pradyot Ranjan <99216956+pradyotRanjan@users.noreply.github.com>

Author: prady0t

src/array_api_extra/_lib/_testing.py

@@ -124,6 +124,7 @@ def as_numpy_array(array: Array, *, xp: ModuleType) -> np.typing.NDArray[Any]:
         return array.todense()  # type: ignore[attr-defined]  # pyright: ignore[reportAttributeAccessIssue]
 
     if is_torch_namespace(xp):
+        array.resolve_conj()  # type: ignore[attr-defined]  # pyright: ignore[reportAttributeAccessIssue]
         array = to_device(array, "cpu")
     if is_array_api_strict_namespace(xp):
         cpu: Device = xp.Device("CPU_DEVICE")