Fix extents of gridded data

[LSgeo]

HR and LR grids have missing rows and columns, likely from not extrapolating past the last available data points. Read the gridder docs and either extrapolate, or keep the outermost sample points. Alternatively, crop to data (check what resolution that gives us). 128x128 would be acceptable, but 200x200 would be best.

[LSgeo]

Supporting future work, it may be best to grid the full array and extract predicted point samples... for gridding again? In order to support arbitrary line spacing selection and gridding. Perhaps 20 m intervals are enough for a reasonable arbitrary resolution method for exploration, and the dataloader needs to be able to select that scale on-the-fly. It would still require fixing and locking the grid extent.

The noddy data are nominally 20 cell sized, but isotropic resolution instead of the usual line spacing deal. At present, we select every 4th row of pixels, which is 80 m line spacing - reasonable for exploration surveys. For arbitrary super-resolution, we could go all the way to 400 m (20 rows apart, 1/10th of the width of the noddy forward model extent). This would give us:

1. 20 m
2. 40 m
3. 60 m
4. 80 m
5. 100 m
6. ... n. n * 20 m