As NYcarbineer points out, your claim is factually false: Presidents have been reelected with a lower percentage of the electorate than they received when they were elected. In addition to FDR, Grover Cleveland was reelected with a lower percentage of the electorate (though Cleveland was not a sitting president when he was reelected).
More to the point, comparing Obama's projected vote share to his vote share in 2008 is a truly awful metric for predicting an election. It penalizes him for doing well in 2008, and it relies on knowing Obama's vote share, which is essentially deterministic by itself. And even if FDR and Cleveland hadn't won relatively narrow reelections there would be no reason to suppose that such a thing couldn't happen.
Of much more predictive power are the swing-state polls you cite. They actually have Obama up
in the swing states. Is the champagne you suggest Republicans drink meant to console them?