if response (y) var is missing from data, error message is late and uninformative

[jo-tham]

after some time without using this; i forgot the idiosyncracy that our default y var name does not correspond to out data

the error message that we get in this scenario (i.e. the --yvar is not in the data) occurs late and would be tricky for users to discern the nature of the error (4 people in yesterday's demo missed it).

run varsel using default value for --yvar

learn varsel --xy-data ./tests/integration/input/data_xy.csv --config ./tests/integration/input/vsel_xy_config.csv

/opt/learn/pysite/lib/python2.7/site-packages/rpy2/rinterface/__init__.py:185: RRuntimeWarning: Error in `[.data.frame`(xy, classifiedVarName) :
  undefined columns selected

  warnings.warn(x, RRuntimeWarning)
RRuntimeError: 'Error in `[.data.frame`(xy, classifiedVarName) : \n  undefined columns selected\n'

pylearn.log

root@2996228a1857:/opt/learn# cat pylearn.log
INFO:pylearn:*** Starting Variable Selection ***
INFO:pylearn:Initial variable count: 708
INFO:pylearn:Identify and organize variable sets (varsets)
ERROR:root:Error in `[.data.frame`(xy, classifiedVarName) :
  undefined columns selected

INFO:pylearn:Copying results to /data/output
INFO:pylearn:*** Starting Variable Selection ***
INFO:pylearn:Initial variable count: 708
INFO:pylearn:Identify and organize variable sets (varsets)
ERROR:root:Error in `[.data.frame`(xy, classifiedVarName) :
  undefined columns selected

INFO:pylearn:Copying results to ./