A difficult example for SCS by Priest.
x = accsum_priestx ( )
a 6-by-1, matrix of doubles
"On properties of floating point arithmetics: numerical stability and the cost of accurate computations" Douglas Priest, 1992 p62 "What is the smallest n for which simply compensated summation is not guaranteed to yield a result with a small forward relative error in an arithmetic which conforms to the IEEE 754 standard. The smallest known to the author is ... Then the exact sum is 2, but the sum computed by the SCS is 3."