Reproducibility package: Machine Learning for Chaos Engineering (Taxonomy and Three-Phase Empirical Evaluation)

Code and raw data for the empirical study accompanying "Machine Learning for Chaos Engineering: A Taxonomy and a Three-Phase Empirical Evaluation of Fault Selection, Hypothesis Generation, and Impact Detection" (in preparation for the Journal of Systems and Software). Includes: infrastructure code, orchestration scripts, statistical analysis, and the paper's full 30-paper literature-taxonomy corpus table with search provenance (chaos-benchmark-code.zip, see analysis/taxonomy-corpus.csv); raw per-run experiment data for the tool benchmark and overhead decomposition (chaos-benchmark-data-bench-a.zip, chaos-benchmark-data-bench-b.zip); and the fault-selection campaign and hypothesis-generation data (chaos-benchmark-data-ml.zip). See analysis/PREREGISTRATION.md in the code archive for the full pre-registered analysis plan.