Add nuclear processes in AdePT

[SeverinDiederichs]

This PR adds the nuclear processes in AdePT.

This PR introduces a LeakStatus, which describes for which reason a particle is leaked (e.g., because it hit GammaNuclear or because it left the GPU region). Then, in the Geant4IntegrationLayer, when leaked tracks are handled and given back to Geant4, the nuclear processes are invoked.
Therefore, the AdePTGeant4Integration layer now holds a pointer (not-owned) to the specialized G4HepEmTrackingManager, that is a member of the AdePTTrackingManager. The G4HepEmTrackingManager holds a pointer to the thread-local hadronic processes. Since there is only a single AdePTTransport instance in async AdePT, each thread must initialize its own Integration layer (this needed to be outsourced), such that it points to the correct G4HepEmTrackingManager (with the corresponding pointer to the thread-local hadronic process).

Physics validation for async AdePT:

Physics validation for sync AdePT:

Performance measurement using 32 threads for 2000 events à 100 e- with 10 GeV being shot at testEm3:

Mode	with nuclear	without nuclear	cost (%)
Sync AdePT	233.403 s	218.304	-6.92%
Async AdePT	163.869 s	163.406	-0.28%
G4HepEm	377.132 s	377.184	+0.01%

Note: Cannot be merged until we upgrade the CI to the latest version of G4HepEm, CI will fail at the moment.

To Do:

• Run high statistics validation for sync and async (1e7)
• Update CI baseline file to include nuclear processes
• performance measurements
• do validation when not using nuclear processes in physics list
• update CI to use latest G4HepEm, after that one is tagged (all additions are added in G4HepEm, just waiting for the tag)

[phsft-bot]

Can one of the admins verify this patch?

[agheata]

Nice, thanks for this important feature!