Update geodesic routines

**Update geodesic routines**

Many moons ago (2012-12-21, to be precise), Paul Wessel had expressed
some interest in improving the calculation of geodesics in GMT 6.
Perhaps, it's time to revisit this.

GMT already depends (indirectly) on PROJ (via GDAL) and PROJ since
version 5.0 incorporates my geodesic algorithms. So it would be
relatively straightforward for GMT to use these. The benefits would be

* Distance calculation always converges
* Accurate to round-off
* Interface for computing way-points efficiently
* Accurate computation of ellipsoidal area of polygons
* Access to differential quantities (might be useful in computing buffers)

**Are you willing to help implement and maintain this feature?**

I'm not a stakeholder in GMT so it would probably be best if someone
else make the code changes. However, I'm happy to pitch in if necessary.

If you can make PROJ 5.0 or later a requirement, that you would be able
to *remove* code from GMT.