refactors, add normalized cost functions to progressive n-back, and a new demo for progressive n-back.

add skewed n-back demo using expyriment.

restructure python codes, and init python file for a new progressive optimizer that controls lumpiness and optimizes in chunks instead of one trial per round.