Reproducibility supplement for fractalfract-4623385 (major revision)

Scripts: run in this order from the unpacked directory:
1. python reproducible_cycle_audit.py
   outputs exact-matrix figures Fig04, Fig15 and Fig19, plus an optional
   Fig18_matrix_only figure (not manuscript Fig18), numerical_audit.csv,
   common_norm_certificate.csv.
2. python reproduce_nonlinear_and_networks.py
   reproduces nonlinear 40-degree-class Caputo PECE simulations with
   mandatory watchdog, seeded 300 synthetic ER/WS/BA graphs, manuscript
   Figures 2,3,5-14,16-18,21 and machine-readable CSVs.
3. python reproduce_beijing_figure.py
   recreates Figure 22 and annual descriptive totals from the previously
   supplied Beijing source CSV (192 rows, 2005-2020). The December 2008
   AIDS count is missing; annual AIDS 2008 is a sum of 11 observed months.

Main manuscript Figures 1 and 20 are conceptual diagrams.

Interpretation: complete-cycle finite-family contraction is not verified
for the 300 synthetic graph family, and recovered case counts do not
identify alpha, network edges or ART-campaign effectiveness.

Data provenance: Beijing data previously supplied by the authors,
credited in the manuscript to the Public Health Science Data Center
of the Chinese CDC. The corresponding author must check applicable
redistribution permissions before public supplemental deposition.
