* `f16v4sum $aDst0:Dst0+1, $aSrc0:Src0+1`

*Half-precision* 4-element vector 2x2 lateral summation to 2-element
*single-precision* vector.