powheg_1 test broken

[Juergen Reuter]

With the NAG compiler and default flags: -C=all -nan -gline -w -f2008 the powheg_1 test fails:

Runtime Error: *** Arithmetic exception: Integer divide by zero - aborting
particles.f90, line 1026: Error occurred in PARTICLES:PARTICLE_SET_BUILD_RADIATION
powheg_matching.f90, line 1605: Called by POWHEG_MATCHING:POWHEG_MATCHING_BUILD_PARTICLE_SET
powheg_matching.f90, line 1581: Called by POWHEG_MATCHING:POWHEG_MATCHING_GENERATE_EMISSION
powheg_matching.f90, line 1047: Called by POWHEG_MATCHING:POWHEG_MATCHING_BEFORE_SHOWER
shower.f90, line 212: Called by SHOWER:EVT_SHOWER_GENERATE_WEIGHTED
event_transforms.f90, line 190: Called by EVENT_TRANSFORMS:EVT_GENERATE_UNWEIGHTED
events.f90, line 445: Called by EVENTS:EVENT_EVALUATE_TRANSFORMS
events.f90, line 648: Called by EVENTS:EVENT_GENERATE
simulations.f90, line 1544: Called by SIMULATIONS:SIMULATION_GENERATE
commands.f90, line 4423: Called by COMMANDS:CMD_SIMULATE_EXECUTE
commands.f90, line 5747: Called by COMMANDS:COMMAND_LIST_EXECUTE
whizard.f90, line 357: Called by WHIZARD:WHIZARD_PROCESS_STREAM
whizard.f90, line 332: Called by WHIZARD:WHIZARD_PROCESS_FILE
main.f90, line 412: Called by MAIN
Aborted (core dumped)

[Juergen Reuter]

nlo_4 also fails.

[Juergen Reuter]

And also the openloops_1 test fails.

[kilian]

The offending line is in particle_set_build_radiation, an assignment between two objects of type particle_t. This is intrinsic assignment and thus undefined if the polarization component contains dangling pointers. Two solutions possible

1. Define assignment for particle_t objects explicitly
2. Change the internal representation such that no pointers are left.

Since I wanted to do 2. anyway (#772), I suggest that I proceed as planned. After the refactoring, intrinsic assignment should be fine.

[Juergen Reuter]

WK, go on with 2. In the worst case, this means that the Jenkins remains contaminated for a week or so. There will anyway not any patches or alpha versions till mid of April.

[kilian]

Thus, check after completing #772.

[cweiss]

I have implemented a work-around particle assignment routine in order to be able to better test the uncombined event generation with nagfor. This also seems to have fixed above tests.