Bugfix: compute_ll_ratios typos. Add detector accuracy plot, compute methods. Change plot filenaming

Add ability to select whether to include an accuracy computation in make_train_step and make_val_step functions