acts: Bug? in ActsExampleFatrasTGeo event processing with Pythia input

I’ve spotted an issue of Fatras simulation when processing Pythia events.

I generated 50 pp 14 TeV events as:

./ActsExamplePythia8 -n 50 \                                                                                
--output-dir=data/pythia8/pp14_n50 \                                                
--output-csv \                                               
--gen-cms-energy-gev=14000

When doing their propagation with ActsExampleFatrasTGeo most events were processed within seconds or faster but then for a few events it took 15/10 mins to process and after that it gets stuck until I stop it manually:

10:34:47    Sequencer      INFO      finished event 36
10:34:49    Sequencer      INFO      finished event 42
10:34:51    Sequencer      INFO      finished event 15
10:34:52    Sequencer      INFO      finished event 43
10:34:53    Simulation     ERROR     Propagation reached the step count limit of 1000 (did 1000 steps)
10:34:54    FatrasSimula   ERROR     event 44 particle id=(62|0|74|0|0)|pdg=K+|q=1|m=0.49368|p=12.7496 failed to simulate with error PropagatorError:3: Propagation reached the configured maximum number of steps
10:34:54    Sequencer      INFO      finished event 16
10:34:55    Sequencer      INFO      finished event 44
10:34:59    Sequencer      INFO      finished event 45
10:35:03    Sequencer      INFO      finished event 46
10:35:06    Sequencer      INFO      finished event 47
10:35:10    Sequencer      INFO      finished event 48
10:35:14    Sequencer      INFO      finished event 49
10:35:17    Sequencer      INFO      finished event 21
10:35:20    Sequencer      INFO      finished event 22
10:35:24    Sequencer      INFO      finished event 23
10:35:28    Sequencer      INFO      finished event 24
10:50:41    Sequencer      INFO      finished event 13
10:59:40    Sequencer      INFO      finished event 14

Events 13 and 14 look similar to other in terms of size:

While processing events 13 or 14 my hard drive lost 10 GB of space. The issue was also reproduced with a similar dataset – when around 5 events are left to be processed it gets slow/stuck.