GPU and Node efficiency

By integrating with the jobstats caching at Princeton University HPC systems, reportseff is able to pull out more accurate node usage statistics and GPU usage. Reports can include values split by GPUs or nodes.

If jobs have other AdminComment contents, an exception will be raised. Please report these so I can modify the parsing routine to be more flexible.