Augmented streams (which have stacked frames shifted) createn copies of the input. At test time the logit streams are averaged together. However, this is buggy under CTC training, as the blank label can dominate other labels in the averaged stream under CTC. Documented more fully under#193 .

Proposed fix: change test code to not create shifted copies after stacking. Also have proposed changes to training to allow dumping of logit stream during cv pass; will output first encountered stream instead of averaging. Discussion welcome.