New approximation: DR-A

The DR-A approximation analyzes the mean delay of class $i$ jobs in an M/PH/$k$ queue with $m\geq 2$ priority classes by aggregating all the higher priority classes, as in the MK-N approximation. By contrast to MK-N, the job size distribution of the aggregated class is approximated with a PH distribution by matching the first three moments of the distribution. Recently, an extension of MK-N to PH job size distributions (hyperexponential distributions, in particular) is also proposed, independently, by Sleptchenko et. al. [194] (we propose DR-A in [204]).)