On the computation of inclusion regions for partitioned matrices