I disagree with your statement that an aldehyde oxygen atom is a better nucleophile than ROTMS. Redraw with H+ instead of TMS+. In that scenario, oxocarbenium formation might be easier to see.
And yes, the orbitals need to overlap in the transition state leading to the tetrahydropyran ring. Any time bonds break, it is because electrons have been pumped into a corresponding antibonding orbtial. For the reaction we described, the orbital alignment requirement is met in the chair-like transition state. In the E/syn case, the chair-like transition state has the C-Si σ bond perpendicular to the C-C π* orbital. As a result, there is no favorable overlap, and the THP forming step does not occur. Only through the boat-like transition state is the orbital overlap requirement met and the reaction can occur.