Fixed IndexError when category names are missing in .json

If a _config.json does not contain some category name, then an error would appear as shown below.

Traceback (most recent call last):
  File "./dtcontrol/venv/bin/dtcontrol", line 11, in <module>
    load_entry_point('dtcontrol', 'console_scripts', 'dtcontrol')()
  File "./dtcontrol/dtcontrol/cli.py", line 597, in main
    args.func(args)
  File "./dtcontrol/dtcontrol/cli.py", line 521, in core_parser
    suite.benchmark(classifiers)
  File "./dtcontrol/dtcontrol/benchmark_suite.py", line 95, in benchmark
    cell, computed = self.compute_cell(ds, classifier)
  File "./dtcontrol/dtcontrol/benchmark_suite.py", line 133, in compute_cell
    cell = self.train_and_get_cell(dataset, classifier)
  File "./dtcontrol/dtcontrol/benchmark_suite.py", line 158, in train_and_get_cell
    self.save_dot_and_c(classifier, dataset)
  File "./dtcontrol/dtcontrol/benchmark_suite.py", line 180, in save_dot_and_c
    outfile.write(classifier.print_dot(dataset.x_metadata, dataset.y_metadata))
  File "./dtcontrol/dtcontrol/decision_tree/decision_tree.py", line 80, in print_dot
    return self.root.print_dot(x_metadata, y_metadata)
  File "./dtcontrol/dtcontrol/decision_tree/decision_tree.py", line 225, in print_dot
    text = 'digraph {{\n{}\n}}'.format(self._print_dot(0, x_metadata, y_metadata)[1])
  File "./dtcontrol/dtcontrol/decision_tree/decision_tree.py", line 260, in _print_dot
    last_number, child_text = child._print_dot(child_starting_number, x_metadata, y_metadata)
  File "./dtcontrol/dtcontrol/decision_tree/decision_tree.py", line 246, in _print_dot
    label = group[0] if not names else names[group[0]]

This MR checks if the name for the action index exists, if yes, it uses it, if no, it uses the action index itself as the name.