Consider all changes inside a time interval

Currently the BM tool focuses on the last change in a benchmark time series, assuming that only the last change is of interest.

But hanges before the last one could potentially also be of interest, especially if they're more significant (e.g. being of greater magnitude) than the last change. Consequently, we should consider if targeting any change inside a given time interval (say, during the last n days) would make more sense than just targeting the last change. The latter could be an optional restriction.