On the role of the local oscillator intensity in optical homodyne-like tomography