Clarifying default algorithm used for triangulate - Watson or Shewchuk?

[weiji14]

Description of the problem

Two things that popped up while trying to wrap triangulate in PyGMT, see GenericMappingTools/pygmt#731 (comment) for context:

1. The documentation at https://github.com/GenericMappingTools/gmt/blame/6.3.0/doc/rst/source/triangulate.rst#L60-L61 says that the triangulation algorithm used can be determined using gmt triangulate -, but that only prints the short help docs. @meghanrjones found that it should be gmt get GMT_TRIANGULATE instead
2. Just want to confirm that the Watson [1982] algorithm is indeed the default, and not the newer Shewchuk [1996] algorithm? The conda-forge gmt package seems to be using Shewchuk [1996], but I don't see any special flag set in the build recipe at https://github.com/conda-forge/gmt-feedstock/blob/def03480bc8c569e58d94359f8a835016f292883/recipe/build.sh#L1-L51. This may have licensing implications, since the Shewchuk [1996] algorithm is not for commercial use according to https://github.com/GenericMappingTools/gmt/blob/6.3.0/src/TRIANGLE.HOWTO.

System information

• Operating system: Linux
• GMT version (gmt --version): 6.3.0

[PaulWessel]

I think it has to be on the user to switch away from Shewchuck if they are using GMT commercially. We should not cripple GMT with Watson just in case. I think Shewhuck is the default - at least that is what it says when I look at GMT_TRIANGULATE.

[weiji14]

Ok, so just need to reword the documentation to say that Shewchuk is the default. I'll open up a PR for this.