Interview Guides
Given a stream of log lines, implement a function that parses each line and returns rolling performance metrics over the last window_seconds. Each valid log line has the format "<timestamp> <service> <latency_ms> <status>", where timestamp is a non-decreasing integer. Ignore malformed lines. For each processed line, output the current metrics for that line's service: total requests in the window, error count (status >= 500), and average latency rounded down.
1 <= len(logs) <= 10^51 <= window_seconds <= 10^60 <= latency_ms <= 10^6100 <= status <= 599 for valid lines
