Add the ability to alert on changes in performance metrics instead of instead of static or anomoly thresholds.

Currently we alert on performance metrics when the value hits a specified threshold. We would the ability to alert on sudden spikes or changes in a performance statistic.

For example, we would like an alert if disk usage jumps suddenly from 10% to 50% but we don't care if it goes from 50% to 60%

It is my understanding that this would be different than anomaly threshold because these devices will not follow a predictable trend in their performance fluctuations, patterns that are needed for anomaly thresholds to work correctly.