Sliding Window Anomaly Detector

Problem

Given a list of real-valued metrics metrics, an integer window size k, and a non-negative real threshold t, process the stream from left to right. For each index i, once at least k values are available, compute the moving average and standard deviation of the last k values. Mark metrics[i] as anomalous if abs(metrics[i] - mean) > t * std for that current window. Return a list of window averages and a list of anomaly flags for each position where a full window exists.

Constraints

• 1 <= len(metrics) <= 10^5
• 1 <= k <= len(metrics)
• 0 <= t <= 10^6
• -10^9 <= metrics[i] <= 10^9