enable nowcasting dataloader to append to the CSV files to create more batches

[JackKelly]

At the moment, the workflow is:

1. Specify the number of batches for each split in the config YAML
2. Run prepare_ml_data.py. Which first creates CSV files specifying the locations of each example.
3. Once the CSV files exist, it'd then possible to re-create batches for each modality. But it's not possible to increase the number of batches requested in the config YAML.

There should be a command line-option like --append_more_examples_to_csv_files which, as the name suggests, increases the number of examples specified in the CSV files, and then creates those new batches for each modality.