Add support for charged molecules in TensorNet (total charge)

[guillemsimeon]

Training using total charges should be possible by just setting the charge hyperparameter to True in the yaml file.

[RaulPPelaez]

There are some error:

FAILED tests/test_model.py::test_torchscript[cpu-tensornet] - RuntimeError: 
Variable 'q' previously had type Optional[Tensor] but is now being assigned to a value of type int
:
  File "/usr/share/miniconda3/envs/torchmd-net/lib/python3.10/site-packages/torchmdnet/models/tensornet.py", line 221
            # I trick the model into thinking that the masked edges pertain to the extra atom
            if q is None:
                q = 0
                ~ <--- HERE
            else:
                batchp = torch.cat((batch, batch[-1] + 1), dim=0)

Default q to something like

q = torch.zeros(1, device=z.device, dtype=pos.dtype)

[RaulPPelaez]

Looks good and GPU tests pass. Lets merge. Thanks @guillemsimeon !