Sensor Data Anomaly

You are tasked with implementing a data structure that processes a continuous stream of sensor readings. Each reading is provided as a pair: a timestamp (an integer representing seconds) and a measurement (a float). The readings will arrive in non-decreasing order of timestamps.

Your data structure must support the following operations:

1. recordReading(timestamp, measurement): Record a new sensor reading.

2. getMovingAverage(duration): Return the average of all measurements within the last duration seconds relative to the most recent reading. The operation should compute the average over all readings with timestamps in the range [current_timestamp - duration, current_timestamp].

3. detectAnomaly(window, threshold): Check the sensor readings for the most recent window seconds to determine if there is any reading that diverges from the moving average of that window by at least threshold standard deviations. If an anomaly is detected, return the timestamp of the first such reading within this window; if no anomalies are found, return null.

Assume that for each window, the standard deviation and average are calculated over the available measurements in that window. The operations should be optimized for efficiency, considering that the recordReading operation may be called frequently.

Implement this data structure and the described operations. Your solution should handle edge cases appropriately (for example, when there are no readings within the specified duration or window).