The NC description for pin 2 is very misleading. This pin is coupled to the internals of the antenna chip whether you like it or not just by virtue of being so close to the conductors inside the chip, so while it may be a DC "no-connect" its still coupled at RF frequencies. My guess about the added stub is that it is primarily there to tune the resonant frequency of the antenna down to the 878 MHz region. It does this by adding extra length to the overall monopole structure, making it electrically longer. Radiation comes from current, and in this configuration, the peak current is likely still near the base (pin 1) of the antenna chip and so the added stub probably isn't changing the radiation pattern all that much.
It is interesting that the tip of the stub is folded back to lie beside the hottest part of the chip radiator, as if to intentionally couple across. This may have some interesting effects on the input impedance, albiet minor, but without simulating this design, its hard to tell exactly how.
It is worth mentioning that the chip is meant to be the radiating element of a monopole antenna. Therefore, it relies on the associated "ground plane" to form the counterpoise. Without this ground plane, radiation would be bad, falling well short of expectations, and input impedance would also be bad. The groundplane is really the "other half" of the dipole.