So we can see whether a discount helps the CTA (clicks) AND the sale (paid
orders), per arm:
1. pw-analytics.js: include the ?code= (the price-test arm) on the campaign-click
Umami event, and fire it when EITHER a utm OR a code is present (coupon links
often carry code with no utm, so those human clicks were being dropped). Also
strip a stray @TrackLink suffix Listmonk glues onto the last query value.
2. coupon_ab_scoreboard.py: one report combining HUMAN clicks (Umami, already
bot-filtered) and PAID conversions (compliance_orders) per arm. Discount arms
map by code->pct (campaign-daily:<date>:<pct> marker); the no-code control arm
is recovered by re-hashing customer_email the same way the builder bucketed it.
Prints clicks, paid orders, revenue, and click->order per arm.
