Because it’s generally too complex to guess the result of two rotations.

Mmmm, "guess"? [:)]

Separately the X and Y calculations are possible. Perform the Y calculation on the result of the X-calculation. You already have the code for both calculations. The only difference is that the Y calculation is done on the in-memory result of the X calculation. Or am I thinking too simple here?
That's basically what I do when needed: set the X angle, export, import, set the Y angle.