Characterization of the Impact of a Deep Learning-Based Time-of-flight Image Enhancement Algorithm on Feature Quantification Using Lesion Insertion and Exam Data from Two PET/CT Systems