Apparently Krugman is one of those people who reserves macro for one part of his brain, and micro for the other, and the two parts do not converse. Here, he seems to forget all the trade theory he knows. A trade deal with South Korea turns out to be a bad thing. The potential increases in exports and imports turn out to be a wash, according to Krugman, and do not increase "demand," so there is no effect on GDP. Further, employment will actually go down, i.e. "demand" does not change, effectively productivity has gone up, so employment must decrease.
Sometimes a little knowledge of national income accounting can be a dangerous thing, apparently. The trade deal with South Korea may indeed have a small effect, but US GDP has to increase. This is just basic gains from trade, which Krugman of course is quite familiar with. As for employment, the effect will work in a manner similar to a productivity increase. There are income and substitution effects, and it could go either way. In any case, we are better off as a nation.
It was a great thing when macroeconomic thought was integrated with what everyone else in economics was doing - in international trade, public finance, industrial organization, and general equilibrium theory, for example. It's too bad that some people want to take us back to the Dark Ages.