feat: use urban typology for activity chain matching

[aburianne]

ENTD (national mobility survey) can be used to generate synthetic population (with Synpop) over any metropolitan french territory. It is mostly used whether there are no available EMD or EDGT (local mobility survey with higer sampling).

ENTD is (main method) filtered by departments (i.e.: Gironde-33, Paris-75) composing the survey area, leaving us with critical low sampling issues. Therefore we aim to change approach on sampling and parameters inputs for agents mobility behaviors.

Our proposal is to keep all of metropolitan France (NO department filter) and to apply a typology based on 4 classes :

• central city
• suburbs
• isolated city
• rural / countryside

Doing so, we assume same mobility behaviors patterns within classes, i.e. : Lyon, Bordeaux, Marseille... and significant deviation between classes.

This 4 classes typology is computed and open source distributed by INSEE. It is refered to as "Unités Urbaines" (UU). INSEE internal main use of this dataset is to adapt statistics methods to context. The data comes as Microsoft Excel (.xlsx) worksheet with a 2010 and a 2020 version with official geographical code milleisim options.

The UU typology being availble both in ENTD and official geographical codes for all of metropolitan France it is suitable for the process.

• review code
• set option name and mecanism ( urban_units or any other relevant name)
• write and validate test (new fake data set of UU and hascodes)
• changes for other territories? (Corisca, Toulon, Lyon, Nantes...)
• write documentation
• do we share the jupyter notebook with ENTD filtering methods comparison for reproductible science, so people can evaluate on any french territory?

[sebhoerl]

• Cleaned up the code
• I propose to name the attributes urban_type and then have either suburb, center_city, isolated_city, but all the remaining zones are none

[sebhoerl]

To use the urban type one needs now to

• set use_urban_type: true in the configuration (this imputes the information)
• set matching_attributes: ["urban_type", "*default*"] this prepends the urban_type to the default matching attributes