VIDAR: Visual-Inertial Dense Alignment and Reconstruction via a Geometric Foundation Model | ArxivCSExplorer