An Easy-to-Use and Robust Approach for the Differentially Private De-Identification of Clinical Textual Documents.