Text this: Dataset for Single Character Detection in Dongba Manuscripts