Use a larger input dataset
requested to merge swenzel/hep-workloads:swenzel/alice-provide-larger-inputfiles into qa-build-alice-digi-reco-core-run3-ma
With this, we should be able to scale the benchmark to 5 timeframes (with 16 events) each.
Therefore with 4 threads, we can have events_per_thread = 4 and up to 5 timeframes.