I'll ignore the 2.4Ghz thing, because while thats what samples are being sent out at, nobody, even Intel, knows what the final clocks will be.
As to the OCing thing, to a degree. For pure potential, the duals might have the speed, but they are on the s1156 platform, leaving them with the same PCI-E issues as lynnfields are seeing. Also, to a lot of people dual cores just aren't enough anymore, leaving the 32nm field open to gulftown or nothing.
I don't particularly like the socket split, but unlike many people I understand the reasons behind it. For sure, some of the consequences do just plain suck.
Early samples are often clocked much lower than the actual retail CPUs are. The final round of engineering samples will be more representative of the final retail product. In any case with clock speeds approaching 4.6GHz on air I don't see why Intel wouldn't release 3.2GHz or 3.33GHz versions of Gulftown unless the power consumption was just out of hand at those clock speeds. (Which I don't believe this to be the case at all.)