feat(pipeline) : Ensure the zone_diffusion_codes for DROM/COM

There were cases where we did have the zone_diffusion_code set to "97"
which can't be right, as it would mean that a given service coyuld be
available across the oceans.

Let's fix it and set the correct, 3-digit department number.

This will also enable their search as we now (since the "new" communes)
search for a match agains commune.departement, which can be 3 digits.

I would love to add a data test for this, but not sure how useful that
is. So far, CHECK_SERVICE_ERRORS() is only semantic but does not
validate for instance that the zone_diffusion_code is a correct
department number if the zone_diffusion_type is "department", and so on.

If we wanted to validate this, what should be the correct way to
proceed?

pipeline/dbt/models/intermediate/int__union_services__enhanced.sql

@@ -6,34 +6,40 @@ structures AS (
     SELECT * FROM {{ ref('int__union_structures__enhanced') }}
 ),
 
-adresses AS (
-    SELECT * FROM {{ ref('int__union_adresses__enhanced') }}
-),
-
 departements AS (
     SELECT * FROM {{ ref('stg_decoupage_administratif__departements') }}
 ),
 
--- TODO: Refactoring needed to be able to do geocoding per source and then use the result in the mapping
+adresses_with_code_departement AS (
+    SELECT
+        {{ dbt_utils.star(from=ref('int__union_adresses__enhanced'), relation_alias='adresses') }},
+        CASE
+            WHEN LEFT(adresses.code_insee, 2) = '97' THEN LEFT(adresses.code_insee, 3)
+            ELSE LEFT(adresses.code_insee, 2)
+        END AS "code_departement"
+    FROM {{ ref("int__union_adresses__enhanced") }} AS adresses
+),
+
 services_with_zone_diffusion AS (
     SELECT
         {{ dbt_utils.star(from=ref('int__union_services'), relation_alias='services', except=["zone_diffusion_code", "zone_diffusion_nom"]) }},
         CASE
-            WHEN services.source = ANY(ARRAY['monenfant', 'soliguide']) THEN adresses.code_insee
-            WHEN services.source = ANY(ARRAY['reseau-alpha', 'action-logement']) THEN LEFT(adresses.code_insee, 2)
+            WHEN services.zone_diffusion_type = 'commune' THEN COALESCE(adresses.code_insee, services.zone_diffusion_code)
+            WHEN services.zone_diffusion_type = 'departement' THEN COALESCE(adresses.code_departement, services.zone_diffusion_code)
             ELSE services.zone_diffusion_code
         END AS "zone_diffusion_code",
         CASE
-            WHEN services.source = ANY(ARRAY['monenfant', 'soliguide']) THEN adresses.commune
-            WHEN services.source = ANY(ARRAY['reseau-alpha', 'action-logement']) THEN (SELECT departements."nom" FROM departements WHERE departements."code" = LEFT(adresses.code_insee, 2))
-            WHEN services.source = 'mediation-numerique' THEN (SELECT departements."nom" FROM departements WHERE departements."code" = services.zone_diffusion_code)
+            WHEN services.zone_diffusion_type = 'commune' THEN COALESCE(adresses.commune, services.zone_diffusion_nom)
+            WHEN services.zone_diffusion_type = 'departement' THEN COALESCE(departements.nom, services.zone_diffusion_nom)
             ELSE services.zone_diffusion_nom
         END AS "zone_diffusion_nom"
-    FROM
-        services
-    LEFT JOIN adresses ON services._di_adresse_surrogate_id = adresses._di_surrogate_id
+    FROM services
+    LEFT JOIN adresses_with_code_departement AS adresses
+        ON services._di_adresse_surrogate_id = adresses._di_surrogate_id
+    LEFT JOIN departements ON adresses.code_departement = departements.code
 ),
 
+
 services_with_valid_structure AS (
     SELECT services_with_zone_diffusion.*
     FROM services_with_zone_diffusion
@@ -96,7 +102,8 @@ final AS (
         adresses.result_score       AS "_di_geocodage_score"
     FROM
         valid_services
-    LEFT JOIN adresses ON valid_services._di_adresse_surrogate_id = adresses._di_surrogate_id
+    LEFT JOIN adresses_with_code_departement AS adresses
+        ON valid_services._di_adresse_surrogate_id = adresses._di_surrogate_id
 )
 
 SELECT * FROM final