Reconcile a print shop's job tickets against its printers' meter readings and work out takings

# Reconcile a print shop's job tickets against its printers' meter readings and work out takings Inputs: `printers.csv` (printer_id, kind, price_per_page), `accounts.csv` (account_id, name), `jobs.csv` (job_id, date, printer_id, account_id, pages, copies, status), `meters.csv` (printer_id, date, open_count, close_count) and `period.json` (period_start, period_end, both dates included). All printers, accounts and jobs are synthetic; prices are in dollars. These rules are the shop's; where `counter-notes.txt` disagrees, the rules win. Rules: - Meter readings: a printer-day's meter impressions are close_count minus open_count. The counter has six digits and rolls over from 999999 to 000000, so when close_count is lower than open_count, add 1000000. - Job impressions: a job's impressions are pages × copies. Jobs with status `done` or `reprint` were printed; `void` jobs were cancelled before printing and count nowhere. - Discrepancies: each printer-day may have up to 15 meter impressions that no job explains (warm-up and test pages). A printer-day is a discrepancy when its meter impressions minus the impressions of the printed jobs on that printer that day is more than 15 or less than 0. - Billing: a `done` job is billed pages × copies × its printer's price_per_page, less 10% when copies is 50 or more (a bulk job). `reprint` jobs redo the shop's own mistakes and are free; `void` jobs are not billed. Don't round until the output. - Write `out/reconciliation.json`: `{"discrepancies": [<sorted "<printer_id>@<YYYY-MM-DD>" strings>], "meter_impressions_total": <number>, "bulk_jobs": [<sorted job_ids billed with the bulk discount>], "billed_total": <dollars>, "takings_by_printer": {"<printer_id>": <dollars>, ...}, "takings_by_account": {"<account_id>": <dollars>, ...}}`. - Roll-ups: Printers in printers.csv with no billed jobs are included at 0. Accounts in accounts.csv with no billed jobs 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.