Add User Guide section on tesseroid layers

[manduinca]

Closes #352

Adds a "Tesseroid layer" section to the tesseroid User Guide page, following the same structure as the prism layer section in prism.rst. It documents tesseroid_layer, the DatasetAccessorTesseroidLayer accessor and its .gravity() method, with a self-contained runnable example over a region in South America.

I kept it as a section inside tesseroid.rst (with a _tesseroid_layer label) rather than a new page, since the prism layer is documented the same way and #352 links to that anchor. No changes to the toctree for the same reason.

The one thing I made sure to highlight in an .. important:: note is the difference from the prism layer: surface and reference must be given as geocentric radii, not heights, so the example uses boule.geocentric_radius to build them.

I ran the example locally and it produces a sensible g_z field. I couldn't build the full HTML docs on my machine (no GMT binary for the PyGMT figure), so I'd appreciate a check that the plot cell renders on CI.

[santisoler]

Thanks @manduinca for opening this PR! I'll go through it and leave a review soon!

[santisoler]

This looks really good @manduinca! Thanks a lot for taking the effort on writing it!

I left two minor suggestions below. Let me know what do you think.

I see that the style check on GitHub Actions are failing, but this is not your fault. I'm fixing it in #699. So don't worry about that!

[santisoler]

Minor fix to the reference to Boule's method:

Suggested change

	We can use the :meth:`boule.geocentric_radius` method from :mod:`boule` to
	We can use the :meth:`boule.Ellipsoid.geocentric_radius` method from :mod:`boule` to

[santisoler]

I think it would be nice to add a plot after this cell with the surface, so readers can see the synthetic topography within the area and also its values, so it might stick better that those are radii and not oscillations around the ellipsoid.

[manduinca]

Thanks for the review @santisoler! I applied your suggestion for the Boule reference and added a plot of the surface right after that cell. The colorbar ends up showing values around 6.37 million meters, so hopefully it's now much clearer that these are radii and not heights. The change of the ellipsoid radius with latitude is also visible in the plot, which I think helps drive the point home. Happy to tweak it if you had something else in mind.

By the way, I just opened #700 with the spherical terrain correction example for the User Guide (#353), which builds on this section. There's a design question in there about how to handle observation points that fall inside the tesseroids, whenever you get a chance.