In the transition of the Histogram filter to the new filter types, a
bug was introduced. The Histogram filter has an option to set the
range for which to compute the histogram. (The range of the data will be
used instead if none is specified.) This range was properly used when
computing the histogram on a PartitionedDataSet. However, when running
on a single DataSet, the new filter structure skipped over the
partition code that checked to see if the range was specified. (It also
skipped over the code that properly computed the range in MPI.)
This change properly checks to see if Execute was called with a single
DataSet and updates the ComputedRange appropriately.