This is more comfortable. Furthermore, it is going to be necessary
when we extend the script to several input SHPC.

For an animation, we want several dates so we need `ishape_last.txt`.

`make_animation` seems too general, the reader may think it is an
object of Matplotlib.

Write `SHPC*` instead of `SHP_triplet_*` directories.

Required since commit f9f99c4.

Since commit 8842ed2, the tests use
`$tests_old_dir/Extraction_eddies_region_4`, which contains two
SHPC. Just use one of them.

Rename `Region_1_SHP_triplet_noise` to `Region_1_SHPC_noise`.

Bug was introduced in commit 8c91ffe5.

Run it with two dates and the new input NetCDF files. Two dates so it
can replace the SHPC that was in `$src_dir/Tests/Eddies_region_4`.

Instead of `$src_dir/Tests/Eddies_region_4`. Because
`$src_dir/Tests/Eddies_region_4` was made with an old version of
`inst_eddies`: cyclones and anticyclones output together.

Since it is a test on script which is in Analysis.

To regroup statements which depend on the shapefiles.

Move `Overlap/Tests/Input/Region_4/Eddies_region_4` to Tests. Now used
both in `Inst_eddies` and `Overlap`.

Move the creation of animation to new script
`anim_eddy_contours.py`. `plot_eddy_contours.py` was becoming too
complex. Some of the options of `plot_eddy_contours.py` did not apply
to the animation. Plus we now want to iterate in
`plot_eddy_contours.py` on cyclones and anticyclones so we are going
to have to add complexity in `plot_eddy_contours.py`.

Using `ishape_last`, instead of iterating through the shapefile.

It does not read, it computes.

Return None from `open_shpc` when `ishape_last.txt` is not found,
instead of `[len(readers["extremum"]) - 1]`, as needed by
`plot_eddy_contours.py`. So that we can take another action in another
script.

So that it is in the grid.