Wrong kwarg name in output modules MLP

[sef43]

Currently using the training argument "output_mlp_num_layers" will not actually do anything with a scalar output model.
The keyword argument here needs to be changed to "num_hidden_layers":

torchmd-net/torchmdnet/models/output_modules.py

Line 79 in f6c0c16

num_hidden_layers=kwargs.get("num_layers", 0),

To correspond to the passed value here:

torchmd-net/torchmdnet/models/model.py

Line 130 in f6c0c16

num_hidden_layers=args.get("output_mlp_num_layers", 0),