Add landslide option -S for grdseamount

[PaulWessel]

This feature makes it possible to recreate some of the ancient landslide modeling that was used in J. Smith and Wessel, 2000 on the isostatic consequence of large landslides in oceanic islands. This feature required a lot of work on the input, processing of options and modifiers, and the basic math of handling the many tunable functions and parameters that defines a landslide, including time duration. The man page has been revamped with 5 new figures to explain the purpose and function of the radial slide shape, azimuthal variation within a sector, and the temporal volume fraction curve during the event. I do not expect anyone to really review this PR since it is very extensive, but I don't think it is necessary. I am actively using it for research and thus any issues will quickly be found as I am exploring various situations. As is, it passes all the existing tests, that is, after I fixed a few problems and had to update some of the plots. The current version is thus better in every way. I would want it merged for 6.4 since a publication we are working on will use this and reviewers may wish to have access.

Here are some of the things (some has nothing to do with grdseamount but got fixed along the way):

1. A few references got reformatted to keep everything the same
2. Five new documentation scripts to make it possible to navigate the settings
3. Extensive improvements to grdseamount documentation
4. Fixed bug where the backwards compatible code l (for linear flux rate) was in fact not checked for but instead mixed up in checking the build mode (which is either c or i).
5. Normalized the error function volumetric curve so it goes exactly from 0-1; otherwise the first (and last) volume increments are way too large and very jarring during a movie.
6. Changed a few of the existing scripts to accommodate the above bug fix so we don't need to update too many figures
7. Fixed error in the flux curve plot - it had a factor of 6/2 instead of 3/sqrt(2) inside the error function
8. Fix the seamount types script so that it handles the singularity in the polynomial equation (no way gmtmath can know this). grdseamount computes that same shape via a function that was fixed to handle the singularity (Let grdseamount -Co handle u = -1 #6476).
9. Fix error in -W parsing where it used GMT_IS_CUBE instead of GMT_IS_GRID. Harmless but wrong.
10. Fixed (via separate PRs) several GMT issues that was discovered during the grdseamount work, such as Fix error in setting max half-width in node search #6547, Fix -R parsing check for geographic data #6536, Address bug in NORM operators #6535, Improve clip path for images in -Qi #6523, and Let grdview -N level default to minimum value #6496; there may be others I forgot.

The -S option lets uses specify one or more slides per seamount, each with a different set of parameters and time history. In crossection it looks something like this (but the shape of the slide is tunable):

[joa-quim]

One day I'll have to find time to see if I can use this as source for tsunami modeling.

[PaulWessel]

Yes, that is a good one. I will now try to grow an entire seamount chain in the Pacific.