TODO

Step 2 may fail for very large systems with a bus error.  This appears to be due to memory, and seems to happen
  when running on a blade.  Exiting to the login node, or running with larger memory, seems to resolve this
Explicitly write log name when reporting permission denied errors
Write disk usage stats in title
Directories should be consolidated so that can scan multiple disk volumes more easily (e.g., overnight run)

TODO:
* It would be good to report on disk usage (du -h) for each volume, to update the FSAudit catalog 
* In plot, order users in alphabetical order
* All output of FSAudit_parallel.sh must go to file, so errors can be caught
* quickly find core files
* separate out logs and output data by timestamp


To do for development: test functionality of `src/summarize_fs.R` by calling
directly via `Rscript`.  In particular, 
* evaluate flags `-D bad.fs` to investigate bad data 
* evaluate -g group.by to enable workflows which do bubble plots vs. volume

Another issue is how to have links to the latest data, timestamp notwithstanding

Output directory should be dat/TIMESTAMP or dat/PROJECT, to structure the output directory a bit more
nicely during multiple runs

# Summer 2020

* replace docker call with WUDocker call
* provide mechanisms to discover uncompressed SomaticWrapper runs.  These may include specific Pindel output
  files (_SI, etc)