perfbranchmap: allow to select the filter to benchmark Running the branchmap computation on all filter levels can be expensive. Narrowing the run to some specific filters can speed up benchmarking time when working only on a subset of filter levels.