Respect rate limits to keep integrations stable for all API consumers.
attempt 1: wait 500ms attempt 2: wait 1000ms attempt 3: wait 2000ms attempt 4: wait 4000ms (+ random jitter)