Does 3D in vivo dosimetry provide superior error detection sensitivity for small field sizes in pre-treatment quality assurance compared to conventional 2D gamma analysis?