Upgrade algorithm for finding plate associated with point

See merge request !563

Berta Vilacís authored Jan 13, 2023 and Marcus Mohr committed Jan 24, 2023

General cleanup of functions that were not used anymore due to
the update to point-in-polygon in 3D using the boost library.

Change also to the interal coordinate system to the usage of the
coordinates lon lat. This change is due to the boost library.
It preforms the operations in the surface of a sphere in such
coordinate system.
Also also to improve usability of the code.
The user doesn't have to change the input coordinates to the unit sphere
anymore. Now, palates can be given any xyz, independant of the radius.

The output is still the velocity of the point in xyz in the sphere of
EarthRadius in m/s.

(cherry picked from commit 9bce1d33)

Berta Vilacís authored Jan 12, 2023 and Marcus Mohr committed Jan 24, 2023

It implements the point in polygon in the surface of the sphere
using the libary boost. To do so, it uses the coordinates in
"spherical-equatorial", lon lat as used in geophysics (see boost page).
Thus the plates boundaries are now stored directly with lon lat,
they are not converted to the unit sphere.

It also calculates the distance from the point to the polygon using
the boost library.

This implementation reduces the amount of operations to locate the point
into a plate and removes the dependency of the module plates
to the library CGAL.
It also reduces the amount of points without a plate, almost to none.
The only times we have a point without a plate, is due to the gaps that
the input file provides.

(cherry picked from commit 163da0bc)

Fixes the bug part of issue #196

See merge request !562

CI Changes: remove old gcc+clang add gcc 12 add trilinos tests

See merge request !561

Pimp math formulas in documentation of DGFunction class

See merge request !559

Make Doxygen recognise LaTeX formulas in DGFunction.hpp

See merge request !558

In order to have Doxygen recognise a LaTeX formula as such it needs to be
placed inside \f[ ... \f] or \f$ ... \f$.

Resolve issue #192

See merge request !557

This is the last step to get rid of these for issue #192.

Because of issue 10 in HFG we can currently not generate epsilon forms
for P1 or P2. So we remove the assembly query methods manually.

We should regenerate the forms for consistency once the issue is fixed
in HFG.

Some forms are currently not easily regeneratable with HFG (see
issue 11 in HFG).

For the test of "P1 diffusion, 3D, with blending (HFG)" the tolerance
is increase from 1e-15 to 1.5e-15.

This form was superseded at some point by p2_full_stokes[cc|var_affine_q3]
and is no longer generated by HFG either.

For this commit we regenerated a first bunch of HFG forms without the
assembly query methods.

This commit is step 1 in resolving issue #192. We remove the following
methods
- assemble2D()
- assemble3D()
- assembly2DDefined()
- assembly3DDefined()
from all form classes besides
- the ones generated with HFG
- the base class Form itself
In the latter we implement them as virtual functions instead of being
pure virtual as before.

Once we have regenerated all HFG forms we can also remove the methods
from the Form class itself.

With this commit we also remove the associated checks from the operators
- EdgeDoFOperator
- EdgeDoFToVertexDoFOperator
- VertexToEdgeDoFOperator

Extend PrimitiveID to support childIDs for cells

See merge request !555