G-VOILA: Gaze-Facilitated Information Querying in Daily Scenarios