command-line utility to export data, run segmentation training, and update model with eScriptorium API and slurm

[rlskoeser]

• use eScriptorium API for getting doc & parts as xml
  • download model, create new model
• write methods to download training data
  • correct region block data in output
• run kraken through command line (or python API if easier)
• create or update model in eScriptorium via API after segmentation training completes #29
• clean up temporary data after segmentation training completes #30
• review part apis and kraken arrow - can we convert directly without going to alto + image

assuming local install for now (running on the cluster) and conda env setup not automated (yet)