Flavor sum for 'test' matrix elements

[kilian]

Since test matrix elements are not handled by OMega, it is not possible to use flavor sums, with rather cryptic error message. This should be remedied.

I just found that a good application for constant matrix elements is to compute the unitarity limit for a 2->2 process, in particular at LHC. Might add this as a normalization possibility.

[ohl]

Unit matrix elements can be produced by O'Mega, inverse phasespace volume for massless particles as well. Will need advanced spells for massive particles;)