Tiny gain in efficiency - point offset

[jonnyhuck]

Lines 742-747 of viewshed.cpp use pythagoras then a geometric offset to locate the top left corner of the area of interest, then they convert to coordinates in pixels (rather than units of the map projection).

This could be done more efficiently by simply minusing radius/2 from each coordinate and then converting to pixel coordinates.

This only happens once per calculation, so not a very high priority!