Many updates to pygg

[depristo]

• Update docs to be clearer on how things work
• Massively expand unit testing
• Change the way that data is transferred to R and loaded into R. Now much more explicit API is needed where you say ggsave(dest, plot, data = X) where X is the data you want to transfer to R.
• Removed pygg/gen_cmds.py as this file is essentially a dup of the material in pygg.py
• provided a simple esc() function for use to escape strings
• Many refactored pieces of code:
  • _to_r eg does all of the work translating python objects to R
  • execute_r now does the actual invocation of R and execution of the R script. Checks return value so that we know R execution has succeeded or throws an exception with a meaningful message
• Moved away from map, filter and instead use list/dict comprehensions
• Always load data from a CSV file instead of trying to inline. Cleaner single approach
• Now requires pandas as the conversion from lists of dicts and dict of lists are done by calling pandas.DataFrame() rather than hand-written code here

[sirrice]

OK this is awesome.

TODOs:

• add click back into the requirements,
• make runpygg.py consistent with new API (will create issues for them)