Product sales over time

1-- amazon_seller_product_sales_over_time - the sales equation at line grain, event
2-- based, so returns land on the day they happened.
3--
4-- Grain: day x seller x marketplace x ASIN x SKU x row type. Currency: the order's.
5-- Depends on: staging.amazon_seller.orders, .returns, .fba_returns, .listings
6--
7-- THIS IS THE MODEL THAT MAKES AMAZON DATA BEHAVE. Everything else in the connector
8-- is a current-state report; this one is an event log.
9--
10-- The mistake it exists to prevent: `SUM(item_price)` off the order report, grouped
11-- by purchase_date. It looks right, it ties to Amazon for a fresh month, and then
12-- it drifts - because a refund issued today reduces the day the order was PLACED.
13-- Last quarter's revenue changes after you have reported it, and nobody can explain
14-- why the same query returns two different answers a week apart.
15--
16-- The fix is to stop treating a sale and its reversal as one fact. A row here is a
17-- financial EVENT: an order line produces a SALE row on its purchase date, and if
18-- it comes back it produces a separate RETURN row on the return date. Signs are set
19-- so the two sum correctly with no special casing:
20--
21--   gross_sales    positive on SALE, zero on RETURN
22--   discounts      negative always
23--   returns        zero on SALE, negative on RETURN
24--   net_sales      gross_sales + discounts + returns   <- plain addition
25--
26-- Row type is part of the grain, so a day where a SKU both sells and refunds
27-- produces two rows and each side stays independently auditable. Aggregate away the
28-- row type in BI.
29--
30-- RECONCILING TO core.amazon_seller.sales_over_time. Sum net_sales over the SALE
31-- rows for a closed month and you land within a fraction of a percent of Amazon's
32-- ordered_product_sales. It is not exact and it cannot be - see the README - so
33-- treat a consistent small gap as expected and a growing one as a bug.
34
35WITH sales AS (
36    SELECT
37        o.purchase_date                     AS date,
38        o.amazon_seller,
39        o.marketplace,
40        o.asin,