Total a small coffee roastery's week by origin, profile and lot after re-weighs and a roast-loss rule

# Total a small coffee roastery's week by origin, profile and lot after re-weighs and a roast-loss rule Inputs: `lots.csv` (lot_id, origin, green_kg), `roasts.csv` (batch_id, date, lot_id, profile, green_kg, roasted_kg), `reweighs.csv` (batch_id, roasted_kg) and `period.json` (period_start, period_end, both dates included). The week's records are synthetic: every lot, origin label and batch is invented, and the loss limit and bag size below are this roastery's own house rules. These rules are the roastery's; where `roaster-notes.txt` disagrees, the rules win. Rules: - Re-weighs: a row in reweighs.csv replaces the roasted_kg of the batch it names. When a batch has several re-weigh rows, the last one in the file counts. - Roast loss: a batch's loss is (green_kg − roasted_kg) ÷ green_kg × 100, in percent. A batch whose loss is above 18 is scorched and goes to cold brew; at 18 or below its roasted coffee is sellable. - Origins: a batch counts under the origin of its lot (several lots can share an origin). - Bags: bags are 250 g. For each profile, bags is the number of whole 250 g bags that profile's total sellable roasted coffee fills (round down once per profile, not per batch). - Green stock: a lot's green_left_kg is its green_kg minus the green_kg of every batch roasted from it, scorched batches included. - Roast days: a day's roasted_kg is the total roasted_kg of the batches roasted that day, scorched batches included. - Write `out/roastery.json`: `{"scorched_batches": [<sorted batch_ids>], "sellable_kg_by_origin": {"<origin>": <kg>, ...}, "bags_by_profile": {"<profile>": <whole number>, ...}, "green_left_kg_by_lot": {"<lot_id>": <kg>, ...}, "roasted_kg_by_day": {"<YYYY-MM-DD>": <kg>, ...}}`. `green_left_kg_by_lot` has one entry per row in lots.csv. - Roll-ups: Origins in lots.csv with no sellable coffee are left out. Profiles with no sellable coffee are left out. Days in the period with no roasts are included at 0. ## 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.