feature: basic results analysis for BehaviorSpace

The issue names no files, tests, or entry points; begin by locating BehaviorSpace's existing results and CSV handling. Narrow the broad notes to one agreed first capability, then verify it against a simple one-parameter experiment and the requested plot or export behavior.

some old notes from a 2009 devel meeting on this:

General agreement, I think that we need to make sure that simple,
broad, basic low-threshold functionality is there before we do anything
too extensive or advanced in any particular area (within this general
area of results collection and analysis). currently it's quite a pain
to do even the simplest stuff, e.g. familiar examples like varying
the density in Fire and getting a plot of the resulting burn percentages

General agreement, I think, that the right next step is to do what
Forrest proposed, namely build something like Ben Shargel's original
tool, but instead of focusing on "heat maps" (which show two parameters
varying), focus on regular plots (which show one parameter varying).
such a UI can support one parameter experiments directly, and we should
support Paulo's proposed simplified output format. The same UI can
support multiple-parameter experiments as long as we require that you
pick one parameter at a time and fix the others temporarily in order to
get a plot.

Further detail/comments on Forrest's suggestion:

• We should support multiple runs per combination of settings -- this is
  basic. We could just take the mean of the runs and show that, probably
  with an error bar or a mustache box or whatever it's called, or maybe
  just plot a dot for each run and let the user eyeball the dots.
• Once you have a plot there should always be an export button to export
  just the data you are seeing Paulo-style.
• It would be awesome if you could start examining the results as they
  come in, rather than having to wait until the entire experiment is over.
  We should probably plan for this from the start.
• You will need be able to open a previously saved csv file and browse
  it using this UI.
• It would be awesome, and maybe not that hard, if you could open
  multiple previously saved csv files and browse the combined results from
  all of them. That would save cluster people the trouble of merging their
  results files themselves.

Other notes:

• Josh U recommended we look at TinkerPlots ("it kind of blew my mind
  for ease of use")
• Michelle says working in this space would be easier if she/we didn't
  have to parse csv files and suck data out of plots. some kind of objects
  holding the results or events notifying of results as they happen (in
  NetLogoListener or whatever)
• Seth said maybe "metrics" should be something that we
  explicitly represent so they can be shared by plots, monitors,
  behaviorspace, etc? currently the solution is to make a global variable
  and/or reporter procedure. but then you have to put that in the
  procedures tab. is that good or bad? this overlaps with the Two Joshes
  plotting work. maybe it's fine the way it is. or maybe some things
  could be made more convenient without making metrics separate and
  explicit, for example, when making an experiment setup you could choose
  from a list of reporters taken from monitors and plots.