Add geospatial spatial-autocorrelation review assistant

[AlonePenguin]

@algora-pbc /claim #16

Summary

• Add a self-contained geospatial-spatial-autocorrelation-assistant/ module for the AI-Powered Research Assistant Suite.
• Review synthetic geospatial manuscript packets for CRS/projection evidence, sensitive-location precision, spatial train/test leakage, high Moran autocorrelation with random splits, full-dataset preprocessing leakage, test-set tuning, covariate provenance, external spatial validation, and reproducibility artifacts.
• Emit reviewer-ready decisions, reproducibility scores, remediation actions, research-gap opportunities, JSON/Markdown/SVG reports, and a short H.264 MP4 demo artifact.

Non-overlap

This is distinct from existing #16 assistant slices and from the separate #17 geospatial sample-provenance guard. This module focuses on manuscript-method peer review for spatial autocorrelation, blocked validation, coordinate/projection evidence, and geography-aware reproducibility before AI review output is released.

Validation

• npm run check
• npm test
• npm run demo
• npm run verify-video -> h264, 960x540, 4.000000s
• git diff --check -- geospatial-spatial-autocorrelation-assistant
• git diff --cached --check

Attempt registered in issue #16: #16 (comment)

Synthetic data only. No credentials, private research data, external APIs, payout systems, or financial accounts are used.