Add json conversion of NWChem output data to taskflow

[chetnieter]

Following issues remain:

• [x] Since the relevant data is in the standard output of NWChem. This data lives in a file whose name is dependent on the job scheduler. Currently is assume the filename format used by Sun Grid Engine.
  • This was addressed by adding a redirect to the submission script removing the file name's dependency on the job scheduler
• [ ] Consider renaming the json file after it is generated since the conversion script just appends '.json' the output filename.
• [ ] Add the location of the conversion script (NWChemJsonConversion) to cluster configuration. Right now it is hard-coded for ulmus. Create docker image that contains the full software stack to prepare the NWChem output for visualization. This includes:
  • [x] NWChemJsonConversion
  • [ ] Avogadro
• [ ] Host this Docker image on DockerHub. Presumably in the Kitware organization rather than my personal space.
• [ ] Consider pulling the docker image from the cumulus stack and not in ansible provisioning.
• [x] Consider moving the json conversion from the upload_output task to a separate task.
• [ ] Deal with the use of 'private' job module functions.