With any design like this, you are trying to minimize cost, which may mean using the smallest core where the required copper area will still fit. Often, by the time you get to the second primary layer (generally the last winding layer), the intervening windings will make the window for the second primary layer "lumpy", not allowing one to fit as many turns in as with the first layers, where you have a smooth surface for winding. It may be better in that case to allow a few more turns in the first primary layer in order to get the rest of the primary winding to fit in one layer. This will have some impact on the on the coupling (and leakage inductance), but can still result in an acceptable design. What you would like to avoid is a situation where you have 1.5 layers or so for each primary half, which is not an efficient use of the available winding window, and will also result in an uneven surface for the next winding layer, making each succeeding layer progressively more lumpy, and may result in a design where the required windings simply won't fit. Sometimes you don't get everything you want all at the same time for a given design - that's engineering.