Where have I ever said that certain voltages do not exist.
Many places. You often call one just a math manipulation of the "one real" voltage in existence. But just as an example, let's just look further in your post:
So does it matter if we say the source is -V43 when it is really V34.
"when it is really V34" is saying the voltage across the terminals is really only V34. That means that a voltage across the terminals can't really be V43. If V43 really can't be a voltage across the terminals then you are saying V43 really does not exist because we are just looking at V34 the wrong way. Ergo, you are saying that V43 does not exist except as a mathematical manipulation of a voltage that is really only V34.
Now do you see how what you are saying reads?
In both cases I mathematically swapped V34 with V43 and got identical answers, but the two actual physical connections provide different results. So does it matter if we say the source is -V43 when it is really V34.
Ok. So we see what you have done by using the transformer in two different configurations. Now how about addressing the transformer I presented that is connected in only one configuration?
For my generator example, we have the 0? displaced, series combined voltages Van + Vnb = Vab at the single-phase transformer. We also have the two 180? displaced, differentially combined voltages Van - Vbn = Vab at the exact same terminals. Both voltage designations are based on their real physical make-up.
It is not that a voltage between terminals is "really only V34" but that there also "really exists" a voltage between the terminals of V43.
Another demonstration that both really physically exist is given in my open-wye example where I show that both series voltages X1->X2+X3->X4 and X2->X1+X6->X5 are present. That clearly shows that both X1->X2 and X2->X1 really are there as the transformer is connected. They produce real physical results in both directions.
It is not that one or the other of "X2->X1" or "X1->X2" is "really only the other one mathematically manipulated" but that both really are voltages.
I know that an AC waveform has a portion above the 0V point, which we often refer to as the positive half cycle, as well as a portion below, which we call negative. However when performing circuit analysis we look at a 'snapshot in time' RMS value when assigning a direction, and not the individual portions of the AC waveform(s).
Perform the circuit analysis however you want, but don't somehow think that a voltage is "really only" something as used in a series additive direction. It is also true that a voltage can "really be" something when we have a potential difference as well.
Because current Ia and Ib are not flowing at the same time, the phase relationship between them is not readily apparent from your graphics.
Then consider the circuit halves separately. If we looked at the top half, we would define the positive direction and time based on the relationship of the voltage and the direction of current leaving the voltage terminal. The same for the bottom half. The two halves are taking positive pulses at different points in time. Bringing them together does not change that.