The goal is to strictly capture, for every function, everything that the pandas test captures plus a whole slew of other test assertions. Are we doing so? Are we accurately using everything that's in the pandas test base on all of our ported functions?