Output folders in --parallelized too long

[AkiNikolaidis]

I think it would be better to make all the outputs immediately available in the first folder, instead of making them so deeply nested. I realize it's not as flexible a solution. Is there any other way we can keep things organized without such long fodler names?

New- '/group_stability_matrix/_compression_dim_600/_affinity_threshold_0.0/_blocklength_1/cluster_method_ward/dataset_bootstraps'+bootstrap+'timeseries_bootstraps'+bootstrap+'/n_clusters'+cluster + '/_similarity_metric_correlation/group_stability_matrix.npz'

Old- '/workflow_output/group_stability_matrix/group_stability_matrix.npz'

[anibalsolon]

We can tweak the DataSink to have this organization for both parallelized and non-parallelized.