Invoice a bakery's wholesale cafes from revised orders, dated price changes, returns and a tiered discount

# Invoice a bakery's wholesale cafe customers for November Inputs: `orders.csv` (every revision of every order: order_id, rev, cafe, delivery_date, item, qty), `prices.csv` (item prices with the date each takes effect), `returns.csv` (unsold goods sent back against an order), `cafes.json` (each cafe's tier) and `manager-summary.txt`. Every amount is in pounds. The files are the record; where a summary disagrees with them, the files win. Rules: - Only the highest revision of each order_id counts; rev is a whole number, so rev 10 comes after rev 9. A highest revision with qty 0 cancels the order. - Charge each order at the prices.csv row for its item with the latest effective_from on or before the delivery_date. - Credit each return at the unit price its order was charged; returns against cancelled orders get no credit. - Gold cafes get 7.5% off goods minus credits, rounded half-up to the penny; other cafes get no discount. - Amount due = goods - credits - discount. Write `out/invoices.json`: `{"orders": {"<order_id>": <final qty>, ...} (counted orders only, cancelled ones left out), "goods": {"<cafe>": <sum of qty x unit price>, ...}, "credits": {...}, "discount": {...}, "due": {...}, "grand_total": <sum of due>}` with every cafe in `cafes.json` in each cafe map (0 where nothing applies). ## How this run works - Work on your own: nobody will answer questions in this task, and writing to `ask/` counts against the run. - Write only under `out/`. Create `out/` if it is missing. Inputs are synthetic; don't fetch anything from the network. - Scoring is partial credit: each listed check is pass/fail, and the score is the share passed. Numbers are compared to within 0.005 unless the task says otherwise.