Test set leak

After sampling the data to create the test set, those files in the test set are not removed from the training set: https://code.ornl.gov/va8/autoencoder/-/blob/master/train.py#L20

When testing here, I am seeing that files in the test set are in the train set: https://code.ornl.gov/ygx/autoencoder/-/blob/bayesopt/bayes_test.py#L32