Fix bug in PDEOperatorCartesianProd

[jkquigley]

Running operator examples with PyTorch backend throws an error shown below:

Using backend: pytorch
Other supported backends: tensorflow.compat.v1, tensorflow, jax, paddle.
paddle supports more examples now and is recommended.
Compiling model...
'compile' took 0.816340 s

Training model...

Traceback (most recent call last):
  File ".../deepxde/examples/operator/poisson_1d_pideeponet.py", line 61, in <module>
    model.train()
  File ".../deepxde/deepxde/utils/internal.py", line 22, in wrapper
    result = f(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^
  File ".../deepxde/deepxde/model.py", line 673, in train
    self._test(verbose=verbose)
  File ".../deepxde/deepxde/model.py", line 866, in _test
    ) = self._outputs_losses(
        ^^^^^^^^^^^^^^^^^^^^^
  File ".../deepxde/deepxde/model.py", line 579, in _outputs_losses
    outs = outputs_losses(inputs, targets, auxiliary_vars)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File ".../deepxde/deepxde/model.py", line 325, in outputs_losses_train
    return outputs_losses(
           ^^^^^^^^^^^^^^^
  File ".../deepxde/deepxde/model.py", line 316, in outputs_losses
    losses = torch.stack(losses)
             ^^^^^^^^^^^^^^^^^^^
TypeError: expected Tensor as element 1 in argument 0, but got list

This change ensures that the losses variable is a compatible type for torch.stack.

[lululxvi]

Fixed at #1925