Build a sampleM for the monadic interface by wrapping DataFRame.Opera…

[amiri]

…tions.Subset.sample in DataFrame.Monad.modifyM, like filterWhereM

I think I did this correctly. I feel uncomfortable with the lack of tests, though, and I don't know how to test this code. I would appreciate any guidance on how to do that here. I could then write tests for the other monadic functions, too.

[mchav]

Thanks so much @amiri! Agreed we should have tests. I'm trying to figure out how to get good property tests working but this feels like it's a fine change to visually inspect.

You can run the following commands to eyeball the results:

$ ./scripts/repl.sh
ghci> :script dataframe.ghci
dataframe> df <- D.readCsv "./data/housing.csv"
dataframe> D.nRows df
-- some number
dataframe> import DataFrame.Monad
dataframe> import System.Random
dataframe> D.nRows (execFrame df (sampleM (mkStdGen 431) 0.1))
-- Should be 10% of previous number.

[amiri]

OK, that worked, but I used execFrameM instead of just execFrame, and I got 2,079 rows out of 20,640. A tiny bit more than 10 percent.