feat: Implement Phase 4 dbt model restructuring

Create neighbourhood-centric dbt transformation layer:

Staging (5 models):
- stg_toronto__neighbourhoods - Neighbourhood dimension
- stg_toronto__census - Census demographics
- stg_toronto__crime - Crime statistics
- stg_toronto__amenities - Amenity counts
- stg_cmhc__zone_crosswalk - Zone-to-neighbourhood weights

Intermediate (5 models):
- int_neighbourhood__demographics - Combined census with quintiles
- int_neighbourhood__housing - Housing + affordability indicators
- int_neighbourhood__crime_summary - Aggregated crime with YoY
- int_neighbourhood__amenity_scores - Per-capita amenity metrics
- int_rentals__neighbourhood_allocated - CMHC via area weights

Marts (5 models):
- mart_neighbourhood_overview - Composite livability score
- mart_neighbourhood_housing - Affordability index
- mart_neighbourhood_safety - Crime rates per 100K
- mart_neighbourhood_demographics - Income/age indices
- mart_neighbourhood_amenities - Amenity index

Closes #60, #61, #62, #63

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

dbt/models/intermediate/int_neighbourhood__crime_summary.sql

@@ -0,0 +1,81 @@
+-- Intermediate: Aggregated crime by neighbourhood with YoY change
+-- Pivots crime types and calculates year-over-year trends
+-- Grain: One row per neighbourhood per year
+
+with neighbourhoods as (
+    select * from {{ ref('stg_toronto__neighbourhoods') }}
+),
+
+crime as (
+    select * from {{ ref('stg_toronto__crime') }}
+),
+
+-- Aggregate crime types
+crime_by_year as (
+    select
+        neighbourhood_id,
+        crime_year as year,
+        sum(incident_count) as total_incidents,
+        sum(case when crime_type = 'Assault' then incident_count else 0 end) as assault_count,
+        sum(case when crime_type = 'Auto Theft' then incident_count else 0 end) as auto_theft_count,
+        sum(case when crime_type = 'Break and Enter' then incident_count else 0 end) as break_enter_count,
+        sum(case when crime_type = 'Robbery' then incident_count else 0 end) as robbery_count,
+        sum(case when crime_type = 'Theft Over' then incident_count else 0 end) as theft_over_count,
+        sum(case when crime_type = 'Homicide' then incident_count else 0 end) as homicide_count,
+        avg(rate_per_100k) as avg_rate_per_100k
+    from crime
+    group by neighbourhood_id, crime_year
+),
+
+-- Add year-over-year changes
+with_yoy as (
+    select
+        c.*,
+        lag(c.total_incidents, 1) over (
+            partition by c.neighbourhood_id
+            order by c.year
+        ) as prev_year_incidents,
+        round(
+            (c.total_incidents - lag(c.total_incidents, 1) over (
+                partition by c.neighbourhood_id
+                order by c.year
+            ))::numeric /
+            nullif(lag(c.total_incidents, 1) over (
+                partition by c.neighbourhood_id
+                order by c.year
+            ), 0) * 100,
+            2
+        ) as yoy_change_pct
+    from crime_by_year c
+),
+
+crime_summary as (
+    select
+        n.neighbourhood_id,
+        n.neighbourhood_name,
+        n.geometry,
+        n.population,
+
+        w.year,
+        w.total_incidents,
+        w.assault_count,
+        w.auto_theft_count,
+        w.break_enter_count,
+        w.robbery_count,
+        w.theft_over_count,
+        w.homicide_count,
+        w.avg_rate_per_100k,
+        w.yoy_change_pct,
+
+        -- Crime rate per 100K population
+        case
+            when n.population > 0
+            then round(w.total_incidents::numeric / n.population * 100000, 2)
+            else null
+        end as crime_rate_per_100k
+
+    from neighbourhoods n
+    inner join with_yoy w on n.neighbourhood_id = w.neighbourhood_id
+)
+
+select * from crime_summary