Instead of reading it from standard input.

Instead of reading it from a separate file.

The idea is to favor long trajectories.

Preparing for more complex script.

Extract code from the main part of `trajectories.py` into new function
`assign_traj_edge`. Because we are going to make the script more
complex.

This will much more convenient for `cost_function.py` and
`trajectories.py`. Note we then have to create a file
`e_overestim.txt` for `Trajectories/Tests/edgelist_Isic_[12].csv`.

Replace function `to_eddy_agraph` by function
`partition_graph_date`. `partition_graph_date` does not depend on the
pygraphviz graph being created from networkx. We will also use it with
a pygraphviz graph created from graph-tool.

If there are no edges in the graph of segments then there are no cost
values. In this case, in recent versions of graph-tool, when
graph-tool loads the graph, it does not create a graph with the
`cost_function` property, even if the property is declared in the
input graphml file.

So they are related to the date, assuming `e_overestim = 10`.

This carries more information.

This variable was already used in other tests.

Because some tests require results from other tests in this series of
tests.

We now want the default behaviour to be renumber. `SHPC_class` expects
sequential eddy indices from 1.

Instead of using old result from Matlab. And run on cyclones instead
of anticyclones, why not, just one less entry in the namelist.

Because it can be created by analysis scripts.

The corresponding modules are not used in the functions.