[b10000-coder]

What this PR adds

• Step 1: Synthetic e-commerce data generator (Faker, 6 CSV files)
• Step 2: DuckDB loader + dbt staging layer (7 models, 36 tests)
• Step 3: dbt marts — star schema (fact_orders + 4 dimensions)
• Step 4: SCD Type 2 customer dimension with full change history
• Step 5: Streamlit BI dashboard over the marts layer
• Step 6: GitHub Actions CI pipeline (78 tests, ~46s)

Test coverage

PASS=78 WARN=0 ERROR=0

Stack

Python · DuckDB · dbt · Streamlit · Plotly · GitHub Actions