Analyze Multi-GB CSV & Parquet Files in Seconds Locally with Open-Source DuckDB + AI

I have a multi-gigabyte local dataset (`sales_2026_*.csv`) that crashes Excel, and I am querying it locally with open-source **DuckDB**.
Here are the column headers and 3 sample rows:
[Paste headers & 3 sample rows]
Write a DuckDB SQL script that:
1) Safely parses messy strings, currency symbols and timestamps using `read_csv(..., ignore_errors=true)`;
2) Uses DuckDB-native ergonomics (`GROUP BY ALL`, `PIVOT`, `QUALIFY` window functions) to calculate: [your analytical question];
3) Exports the final summary table to `result_summary.csv`.