Types of data items in data object serialized in DataGenerator

@AlexanderFengler Check the boxes with objects to serialize below in data. Feel free to expand the discussion as needed.

• ('cpn_data', <class 'numpy.ndarray'>),
• ('cpn_labels', <class 'numpy.ndarray'>),
• ('cpn_no_omission_data', <class 'numpy.ndarray'>),
• ('cpn_no_omission_labels', <class 'numpy.ndarray'>),
• ('opn_data', <class 'numpy.ndarray'>),
• ('opn_labels', <class 'numpy.ndarray'>),
• ('gonogo_data', <class 'numpy.ndarray'>),
• ('gonogo_labels', <class 'numpy.ndarray'>),
• ('theta', <class 'numpy.ndarray'>),
• ('lan_data', <class 'numpy.ndarray'>),
• ('lan_labels', <class 'numpy.ndarray'>),
• ('binned_128', <class 'numpy.ndarray'>),
• ('binned_256', <class 'numpy.ndarray'>),
• ('generator_config', <class 'dict'>), # nested dict (see below)
• ('model_config', <class 'dict'>) # nested dict (see below)

In data['generator_config']:

• ('output_folder', <class 'str'>),
• ('model', <class 'str'>),
• ('nbins', <class 'int'>),
• ('n_samples', <class 'int'>),
• ('n_parameter_sets', <class 'int'>),
• ('n_parameter_sets_rejected', <class 'int'>),
• ('n_training_samples_by_parameter_set', <class 'int'>),
• ('max_t', <class 'float'>),
• ('delta_t', <class 'float'>),
• ('pickleprotocol', <class 'int'>),
• ('n_cpus', <class 'int'>),
• ('kde_data_mixture_probabilities', <class 'list'>),
• ('simulation_filters', <class 'dict'>), # nested dict (see below)
• ('negative_rt_cutoff', <class 'float'>),
• ('n_subruns', <class 'int'>),
• ('bin_pointwise', <class 'bool'>),
• ('separate_response_channels', <class 'bool'>),
• ('smooth_unif', <class 'bool'>),
• ('kde_displace_t', <class 'bool'>),
• ('dgp_list', <class 'str'>)

In data['model_config']

• ('name', <class 'str'>),
• ('params', <class 'list'>),
• ('param_bounds', <class 'list'>),
• ('boundary_name', <class 'str'>),
• ('boundary', <class 'function'>), # Serialize? If so, how?
• ('n_params', <class 'int'>),
• ('default_params', <class 'list'>),
• ('nchoices', <class 'int'>),
• ('choices', <class 'list'>),
• ('n_particles', <class 'int'>),
• ('simulator', <class '_cython_3_0_12.cython_function_or_method'>), # Serialize? If so, how?
• ('constrained_param_space', <class 'dict'>) # nested dict (see below)

In data['generator_config']['simulation_filters']

• ('mode', <class 'int'>),
• ('choice_cnt', <class 'int'>),
• ('mean_rt', <class 'int'>),
• ('std', <class 'int'>),
• ('mode_cnt_rel', <class 'float'>)

In data['model_config']['constrained_param_space'] [these are tuples of floats]

• ('v', <class 'tuple'>),
• ('a', <class 'tuple'>),
• ('z', <class 'tuple'>),
• ('t', <class 'tuple'>),
• ('theta', <class 'tuple'>)

Start by locating DataGenerator and the code that serializes the data object. Compare its behavior with the listed NumPy arrays, nested configuration values, tuples, and function or Cython callables. Done means the serializable items and treatment of unsupported callable values are decided and reflected in the relevant serialization behavior; no file or test is named in the issue.