emonLib, which is what the emonPi uses for the power measurement and calculations, doesn’t attempt to measure the reactive component. It only computes the average real power, rms voltage and rms current; and it’s from the latter pair that apparent power is calculated. All those values are available, but as you read, only rms voltage and real power are used.
Provided that the power factor remains close to unity, you could use Pythagoras’ Theorem on real and apparent power, but as the power factor gets lower, the errors will increase very rapidly, and in any case you have no way of knowing the sign (capacitive or inductive) of the reactive power.
This emonTx sketch (?) might help. Porting it to the emonPi might be feasible, but there’s every indication that maintaining all the other things that the Atmel 328P does at the same time - particularly handling the r.f. traffic - could well prove to be an insoluble problem.