Using pandas, add a derived column to a table based on multiple conditions with strict precedence and missing-value handling. Given the sample DataFrame below, create a new column 'risk_tier' with the rules: if returns >= 2 OR last_review_rating <= 2.0 then 'high'; else if amount >= 200 AND country in {'US','CA'} then 'medium'; else if signup_date is within the last 30 days relative to 2025-09-01 then 'new'; else 'low'. Requirements: vectorized solution (no Python loops), correct dtype handling for dates and floats, NaNs in last_review_rating should not trigger 'high' unless returns >= 2, and write one simple unit test. Sample DataFrame df: +----------+---------+--------+---------+-------------+-------------------+---------+ | order_id | user_id | amount | country | signup_date | last_review_rating| returns | +----------+---------+--------+---------+-------------+-------------------+---------+ | 1 | 101 | 120.0 | 'US' | '2025-08-15'| 4.5 | 0 | | 2 | 102 | 350.0 | 'CA' | '2025-06-01'| 2.0 | 1 | | 3 | 103 | 50.0 | 'FR' | '2025-08-25'| null | 2 | | 4 | 104 | 500.0 | 'US' | '2025-09-01'| 5.0 | 0 | +----------+---------+--------+---------+-------------+-------------------+---------+