Avoid to recompute pooled travels.

[nicolas-roelandt]

There is no mechanism to check if a pooled travel as already been computed.

For example, the route mixing 2 itineraries is always recompute : https://github.com/Universite-Gustave-Eiffel/geodev-logicout/blob/0edc6e45a8e96066943fcf39bcab88b71fba2ec7/scripts/mutualisation.py#L63C1-L63C47

Same for results from Logicout API

Suggestion of implementation:

• Open ranked_mutualisations.csv, trajets_mutualises.csv and gains.csv as pandas dataframe in main.py
• make antijoins to remove ids from ranked_mutualisations.csv that are already in gains.csvfor examples
• loops on the XX first rows of the rows that are left
• append results to gains.csv| and , trajets_mutualises.csv