Coordinate Deviation When Shifting GeoPoints

[LukasBaecker]

While writing tests for PR #183 I observed an issue with GeoPoints. I compared shifted coordinates (using the shift function of GeoPoints) to points that were calculated with cartesian values and shapely and observed a deviation of two points that should be the same. The difference is very small and as you can see in my tests, when rounded to the 9th decimal digit (which displays 0.11mm) the points are comparable.

Would it be useful to round the coordinates, which are currently saved with 15th digits?