SEH post-review modules P8–P12: specifications, pre-run hashes, code, tests and outputs (shared exploration history in mineral prospectivity validation)

This record contains the post-review modules P8–P12 of the paper "When Spatial Validation Learns Exploration History: Shared Survey Coverage and Discovery Records Inflate Mineral Prospectivity Validation and Overstate Greenfield Targeting Performance" (submitted to Natural Resources Research). The modules were written after external adversarial review rounds to test the gaps the reviews identified. For each module, the specification (decision rules, margins and stated expectations), the code and the unit tests were frozen by SHA-256 before any production seed was run. Freeze times are recorded in the PRE_RUN_HASHES files: P8 2026-10-27T23:31Z; P9 2026-10-28T13:22Z; P10 2026-10-28T14:27Z; P11 2026-10-28T15:24Z; P12 2026-10-28T18:48:23Z. Output hashes are appended below the frozen_utc line after each run. This deposit verifies the integrity of these files; the timing record is the internal hash file of each module. Modules:- P8: budgeted targeting capture (P8A), other learners and gridding including GBM, deep random forest, IDW and ordinary kriging (P8C), DGP sensitivity and explored-ground negatives (P8D), and Nevada clustered thinning and a measured gravity field (P8F).- P9: the sign-reversed measured field in Nevada.- P10: co-located survey masks (P10A), exploration that avoids favourable geology (P10B), and stratified explored-ground negatives (P10D). P10C was not approved and was not run.- P11: validation on independently recorded occurrences (P11A), budgeted capture at realistic settings and on a low-support list (P11B), a Nevada field–history correlation grid (P11C), and clustered (Thomas-type) occurrences (P11D).- P12: greenfield targeting in the standard sense, meaning targets away from known deposits. Also included: the multiplicity tables (holm_v8.py, HOLM_V8_TABLE.md), the code for Fig. 7, a README and SHA256SUMS.txt. P8 run note: the computing environment was reset once during P8F (measured field). The partial rows were deleted and the condition was rerun in full with the same frozen code and seeds. Frozen upstream files (seh_confirm.py, seh_post_p3p6.py, seh_post_p7.py, NEVADA.ISO.txt) are identified by their SHA-256 in the PRE_RUN_HASHES files and are archived in the confirmatory and P3–P7 records.