The way I see it, yes, glass is more important, but it's not really saving you money to spend twice or more trying to save a couple hundred on bodies, especially if major-purchase money doesn't come along often.
Making the camera the weak link for some years isn't really helping you make use of the better glass, necessarily, anyway (Depending what you try to do.) I'd jump for at least either a K-x, K20d, or K-7: as mentioned, glass holds its value better so you won't lose much by trading up as you go. (This is, by the way, in a context where the earlier digital cameras would be making pretty big leaps in capability and such pretty frequently: as the technology matures, the differences between them should matter less and less in more and more situations. 'Good enough' is good enough.
)
The K-5 and K-r are impressive... If I were making a ten year commitment I might indeed go all out there, but I've got a K20d (bought at about the lowest point of new prices for these, but still a big investment, ) on the rationale that this fellow's reached a level where I can do just about everything digitally I could with film, so he constitutes a good hold-point. Basically, since cameras depreciate, but lenses more or less don't, the camera's where you want to please yourself, primarily. It's you, not the market, that's going to appreciate the time between. Laying out a few hundred periodically just to upgrade a camera *is* a lens-sized investment, possibly just to be dissatisfied at any given point.
For me, the K-5 looks like a substantial enough upgrade that if I need/can have one, (Being pretty nocturnal, the high-ISO and AF improvements are a big deal to me) they'll be down to a reasonable price soon enough.
Basically, paying the 'brand new model premium' might be something you could balance against what glass you can start with, but starting off with a camera body you can be happy with a long while actually kind of frees you to improve your glass situation. Basically, there's diminishing returns about going cheaper than you have to.