1.05b

• Fixed tiny but major bug where the Test Classifier was sending testing data to training set folds that contain the same data in cross-validation, resulting in unrealistically accurate predictions. This was caused by a change in the last update, so any exported Test Classifier results that used cross-validation with version 1.05a should be considered invalid.