Map Imputing Zeroes: refactor R util function

[d-callan]

i did a thing here #19 .. and then i got tired and ran out of time and wasnt sure what the future would hold besides and thought, i have to call it somewhere. so im filing this ticket to keep track of some things that i maybe could have done but didnt, and might have to do yet sometime in the future. those things are:

1. make a MegastudyPlot or similar class that has slots for a Megastudy and a VariableMetadata object. move some validation from the imputing method to this class validator fxn. (this could ultimately lead to a reworking of the plot.data API..)
2. moving the helper fxns inside the imputing method to stand on their own and have explicit tests
3. make sure the imputing method never removes columns, or make it so that it always removed columns it doesnt need. (it shouldnt hurt anything that it does remove them, but its done inconsistently)..