Maybe I don't understand the question, but I think the confusion comes from differences in the grids between your two kinds of datasets. The whereami output is on the grid of the Talairach Daemon atlas, and the resampled data is at a coarser resolution of 3mm and offset by 0.5 from the Talairach atlas. The "Jump to xyz " function in AFNI jumps to a coordinate based on the underlay dataset. The 3dinfo output shows the extent coordinates and voxel sizes, and that all seems expected.