Given a list of event records, write a function that returns summary insights per customer. Each record is a list in the form [customer_id, event_type, amount], where customer_id and event_type are strings and amount is a non-negative integer. Aggregate total amount per customer, count the number of events per customer, and return the top k customers sorted by descending total amount, then ascending customer_id for ties.
1 <= len(records) <= 10^5[customer_id, event_type, amount]1 <= len(customer_id), len(event_type) <= 200 <= amount <= 10^61 <= k <= number of distinct customer_id values