Evaluation in Center Of Mass frame vs out of COM.

We noticed (in both ppllj and ppllj_ew) that probing close to a divergence gives rise to NaNs, however boosting the momenta to center of mass we get a result (that is quite large as expected).

We (Geneva) will probably add a fallback for boosting to COM in case OpenLoops gives a NaN, but we also wanted to share this information with you, in case you think openloops should boost to COM instead of the user of OpenLoops.

I attach an example standalone file with two hard coded phase space points, that should evaluate to the same matrix element, but the out of COM gives NaN.

Cheers, APN

test_openloops.cpp