AFAIK it's because they last a long time with very small amounts of reactants, are low maintenance, and are potentially very small. I mean, it kinda makes sense but RTGs are better for space applications because you really don't need massive amounts of power in space, plus the whole "power forever with no effort" aspect; they might as well be perpetual motion engines.
LFTR's produce 239Pu (among a mess of other stuff), that's the stuff that goes BOOM in nuclear bombes, not 238Pu that glows warmly to make power for spacecraft.
As Jobu points out from Wikipedia: he second proliferation resistant feature comes from the fact that LFTRs produce very little plutonium, around 15 kg per gigawatt-year of electricity ... This plutonium is also mostly Pu-238. According the article this seems like quite a bit more than is currently being produced, is it not viable for use in RTGs for some reason? Or is the cost of a LFTR over the course of a year less cost effective than the current method of Np to Pu?
I have very little knowledge of the science here, I'm just not sure what all I'm missing.