integrating kraken

• changes for new release: getting the transcriptions no longer uses 'type'
• run_volumes gets the pages recursively
• works for both kraken and kaldi with the option -m kaldi / -m kraken
• compared between the data directories produced by each branch for kaldi and there are no difference between both (except for the partitions' files because the script randomly splits the data) which confirms that the script works correctly -resolves threads added by @kermorvant