nx is 0 if the script is run on a triplet of shapefiles for
interpolated eddies.

The script can plot more than a snapshot.

The repeat option does not work. Apparently a Matplotlib bug.

Bring plotting commands together at the end of the main part of the script.

Extract instructions to select ishapes from function snapshot to a new
function `select_ishapes`. Preparing to use the function to find the
bounding box for the selected shapes.

Preparing for extraction of selection task into a function.

Especially important for animation.

Cleaner and more powerful.

We can access the projection through the axes instance.

They do not change for a new date, for instance.

Too complex. Just re-run the script for a new figure.

Replace argument `shp_tr_dir` by arguments k1, `ishape_last`,
`reader_extr`, `reader_outer`, `reader_m_s`. The idea is to not redo
tasks in function snapshot when calling it for several dates.

Move argument window before the keyword-only arguments. Preparing for
using the function in an animation.

It is clearer that args.window in the main part of the script and
window in the function snapshot both contain the tuple or None. Also,
function snapshot referred to global variables llcrnrlon, llcrnrlat,
urcrnrlon, urcrnrlat which were defined under the test `__name__ ==
"__main__"`. That was not convenient when importing the script. This
is fixed.

Without the assert, if the current axes is not a Geoaxes instance,
there is a less clear error with a plot instruction later.

Else we loose the zoom on the window.

So k is no longer optional. Preparing for using the function in an animation.

Add options: plot a single component; label the leafs, the splits or
the merges, or all eddies, instead of the roots. Add functions for
easier hacking.

From lines that were in the main part of the script.

When we want to produce a Graphvizfile, detect interpolated eddies in
function `read_eddy_graph` rather than in the main part of the
script. This is longer because we read and set all the meta-data in
extremum, and also because we do an additional loop on nodes of the
graph. But it is better in preparation for a function that will create
a Graphviz file for any subgraph: we do not want to read the
shapefiles again when we call the function several times.

Add argument to function `read_eddy_graph` to separately control the
reading and setting of attributes of interpolated eddies.  This is in
preparation for a transfer of this task from the main part of the
script to function `read_eddy_graph`.

The previous version annotated a single ancestor, and it was not
necessarily the ancestor with the oldest date. Moreover, this revision
is probably quicker: no need for subgraph and topological sort.

Now useless.

Define attribute `coordinates` of nodes instead of separated
dictionary pos. Use function `report_graph.read_eddy_graph` to define
the attributes of nodes. So we do not need the functions
`compute_ishape` and `get_coord` any longer. Note that, for visible
eddies, `plot_traj.py` used to loop on nodes of the graph and access
corresponding shapes, while `report_graph.read_eddy_graph` loops on
the shapes and accesses corresponding nodes. This is probably quicker.

Add `grid_nml.txt` to the triplet of shapefiles for interpolated eddies.

`n_valid == 0` occurs if the script is used on the shapefile triplet
of interpolated eddies.

Not used for now in the scripts.