Interactions are not correctly filled in rescans from HepMC

[Juergen Reuter]

I take the same example as lhef4.sin but with LHE format replaced by HepMC.

[kilian]

Unless someone else does it, probably after #499.

[kilian]

Next item.

[kilian]

This failure is, unfortunately, an indication of an intrinsic deficiency in the reconstruction of an interaction, given an external particle set.

The various event formats require some reshuffling of the particle content. We therefore have to match a particle set to WHIZARD's interaction without knowing the correct permutation. This applies even for event files that WHIZARD itself has written, although there the possibilities are limited. In any case, I have to implement a more generic algorithm.

[kilian]

The new event reader (actually, transformation from particle_set to interaction_t) takes into account parent-child relations, duplicate entries, and allows for reshuffling the outgoing partons. This should cover a wider range of possible event file setups than before. The minimum requirement is that the incoming partons of the hard interaction are marked as such.

This fixes not just rescanning from HepMC, but also other formats. There is a bug left connected to recalculating the matrix element, #702.