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

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

No need since we want all the variables in the file.

Bug fix: take into account ghost eddies in global Matlab adt
files. Some eddies are duplicated in global adt files, at the border
of the longitude domain. We should not use these ghost eddies. So
R\'emi has added variables `id_anti_cell` and `id_cyclo_cell` in adt
files. Read them. They should contain numbers from 1 to the number of
eddies, with a step of 1, intersparsed with NaN values. Check that
when reading the variables. Add an argument `cell_mask` to function
`write_to_shp` that signals ghost eddies.

Bug fix: add default values for arguments of function
`write_to_shp`. This fixes the case when the adt file is missing. The
bug comes from commit 28afccf5. For simplicity, we remove the empty
list actual argument for cell, in the case of a missing file. Then we
have to test in `write_to_shp` whether cell is None.

Move argument cyclone of function `write_to_shp` before argument
cell. cyclone does not depend on `matlab_data`. We are preparing a bug
fix for the case of a missing adt file.

The name write is too common.

There may be four shapefiles with the centroid layer.

Instead of using values of global variable `matlab_data` inside the
function write. This is clearer. Especially since we already had an
argument of write, cell, which is a value of `matlab_data`.

Integrating work done by Mevludin Isic. We add the possibility to
split `id_child` because it takes more than 2 GiB for the global
domain and thus we cannot save the whole variable to Matlab v6
format. We split `id_child` along the time dimension. As the Matlab
code for this is the same for anticylonic and cyclonic eddies, we
extract it from the script `overlap.m` to a new script
`id_child.m`.

So we can easily import the functions in the file.

Rename component `list_vis` of derived type snapshot to list. Since
there are no longer invisible (interpolated) eddies.

Remove test on orientation in `candidate_overlap`. This is useless
since commit 70e88286. We process an SHPC containing a single orientation.

In commit d5e89eb9, the test was added with the useless key
"input". Now, after commit 29c37646, the useless key is the key "required".

Now that `inst_eddies` reads `main_nml` only on the first date, we do
not need to create a file `main_nml.txt` before running
`inst_eddies.py`. `inst_eddies.py` can just read `main_nml` from
standard input.

If we are appending to an existing SHPC then read `main_nml` from file
`used_nml.txt` instead of from standard input, and do not re-create
file `used_nml.txt`. It would be a mess to have an SHPC made with
different run parameters for different parts of the shapefiles.

Preparing to add exist as an argument of config.

Move the call to `find_package` for EZMPI to
`Overlap/CMakeLists.txt`. We are reverting in part commit f04aea8b
since `test_set_max_speed` no longer uses MPI.

Thanks to commit 5ff9e1d2, we are reverting in part commit f04aea8b.

Rename component `number_vis_extr` of derived type snapshot to
`number_extr`. Since there is no longer another compoment including
interpolated eddies.

Replace `SHPC_(anti|cyclo)_all_dates` by `SHPC_(anti|cyclo)`.

Otherwise the Fortran executable itself appends to `ishape_last.txt`.

(Writing times when the Fortran executable did not run was justified
by the existence of `elapsed_cat`.)

Create `SHPC_{orient}` before the loop on dates.

Since the program `inst_eddies` now directly appends to shapefiles.