HiOS verification supplement for manuscript v0.7

This archive retains the v0.6 component code and recorded outputs without alteration and adds a scenario inventory reaggregated during the manuscript review. The device experiment remains 27 selected cases and 81 diagnostic-policy traces. These are consistency tests using shared fixture admission predicates, not real-model or physical-device measurements. Expected labels are loaded before each matrix run; independence from policy-development iterations is not documented.

The experiments/ directory is identical to the previously verified v0.6 archive. Historical README references to local release scope describe that earlier preparation state. This v0.7 supplement is prepared for distribution with the research preprint as ancillary material. The inherited Swift MIT notice remains applicable to those files; no additional general-purpose reuse license is asserted here for the new research files. No private application history, account credentials, or real user data are included.

Device reproduction from the archive root:
python3 -m unittest discover -s experiments/device_contract -v
python3 experiments/device_contract/runtime.py --out device-reproduction
python3 experiments/device_contract/analyze.py --out device-reproduction
python3 experiments/device_contract/render.py --results device-reproduction --table device-reproduction/table.tex

See the paper and device-scenario-facts.json for full scope and metric definitions. The original retained profile_rebindings field includes initial non-default bindings; the new inventory separates initial bindings from later rebindings.

The inventory can be regenerated at the archive root with python3 extract_device_scenarios.py. This portable helper reaggregates saved records and does not rerun policies.
