Revisiting the Experimental Design Choices for Approaches for the Automated Retrieval of Duplicate Issue Reports