Following updates of the libraries.

The programs in this project do not depend directly on them. If they
are secondary dependencies, it is the responsibilty of the primary
dependencies to find them.

Necessary since commit 6e52178b.

Necessary since commit 6e52178b.

Necessary since commit d385d60c.

Rename variable id_child to part_id_child inside the mat files. This
is necessary since commit 6e52178b.

Following updates of the libraries.

Before input of `id_child` because reading `N_eddies` and `date_num` is
very quick, while reading `id_child` is long. So we do not waste time
if `N_eddies.mat` or `date_num.mat` are missing.

The name of the script cannot be the same as the name of a variable in
the script.

The name of the script cannot be the same as the name of a variable in
the script.

This is now a small part of the conversion to v6 for `overlap_v6.py`,
the other part being in `id_child.m`.

Only three files were left in cmake, not enough to justify a directory.

Change of philosophy. We do not force the choice of compiler options
from the project, we leave them to the user at configuration. We just
need to display a warning for the option `-ffpe-trap=invalid`.

This is nicer for the user than waiting for the error because
`my_date` is undefined .

Because the node indices do not start at 1 for the second date.

Preparing the addition of an additional test file, which justifies a
new directory.

Variables `id_anti_cell` and `id_cyclo_cell` contain node indices, not
eddy indices. So we check that the first value is consistent with d,
`d_init` and `e_overestim`. To decide whether we need to read
`node_id_param.json` and filter ghost eddies, we need to look into the
first Matlab file whether there is a variable `id_cyclo_cell`. So we
modify the structure of the script a little: we process the first
Matlab file before the time loop.

Clearer, instead of accessing global variable d.

So we can use the script on an SHPC converted from Matlab.

An SHPC now contains a single orientation (since after v0.12).