Add option for unlabelled dataset in YAML

[andreassot10]

Currently, the YAML uses the actual data to create the predictions table (mod_predictions_unlabelled_data). This is for demonstration purposes.

This dataset will need to be changed to a data frame of truly unlabelled data when these become available.

Meanwhile, add a dataset_unlabelled in YAML and set x = get(get_golem_config("dataset_unlabelled")) in mod_predictions_unlabelled_data_server. This will offer the user the option to specify the unlabelled dataset. Until then, I will keep using the current, labelled, dataset as if it was unlabelled, i.e. dataset_unlabelled: text_data in YAML.