The motivation is to be able to create the files `*traj*.json` for
only a component of the graph, so that `plot_traj.py` can plot the
trajectories in a single component, with different colors for
different trajectories.

To function `plot_single_traj`.

To function `plot_single_traj`.

Instead of storing it in the repository.

This should have been removed in commit 311c0598.

Remove timings from inside function `create_traj_list`.

Move function `create_traj_list` to a separate file.

Preparing for an independent script.

This should have been changed in commit 122b78b6.

Not worth it any longer. The function was introduced in commit
480800c9, when it had more to do.

Separate the construction of `traj_vert_ind` and the definition of
`traj_prop`. The motivation is to be able to create the files
`*traj*.json` for only a component of the graph, so that
`plot_traj.py` can plot the trajectories in a single component, with
different colors for different trajectories. A side benefit is that
the script `trajectories.py` becomes clearer. I have measured on the
global graph for 1993-2023 that this modification does not make a
significant change on the running time of the script.

This is useful to go from vertex to trajectory. For example, Artemis
wants to find the duration of the trajectory before each merging.

We want to be able to re-compute the cost, starting from the output of
a previous run of `cost_function.py`, when we have discarded the input
file of the previous run.

Segments is no longer a test with only isolated segments since commit
e1c18ad8.

Add tests `Cost_function_avg_date`, `Greece_date` and
`Eurec4A_cost_date`.

Add option `--avg_fix` in tests to reproduce previous behaviour.

The old behaviour: averaging over a maximum number of instantaneous
eddies, becomes an option.

We do not use bisect because `max_delta` should be small, around 6,
and most often there should be only one time step between
instantaneous eddies, so the first guess at distance `max_delta`
should be close to the result. Moreover, we would need to define a
date function if we used bisect.

Preparing for generalization of these functions.

This does not change any result, it is just an equivalent
formulation. We are preparing for a generalization of the choice of
`ip_beg` and `ip_end`.