Capture neuro-symbolic cognition role boundaries

[mdheller]

Summary

Captures the ASU / CHRONOS neuro-symbolic integration at the Superconscious cognition-loop layer.

This PR keeps Superconscious in its correct role: visible governed cognition coordinator. It may request, classify, record, and emit safe traces for neuro-symbolic work; it does not authorize policy, execution, routing, schema, ontology, model governance, memory promotion, or runtime action.

Changes

• Add docs/neurosymbolic-cognition-position.md
• Add docs/neurosymbolic-capability-role-matrix.md
• Add docs/neurosymbolic-failure-modes.md
• Add examples/neurosymbolic-cognition-run/task.json
• Add tests/test_neurosymbolic_boundary_docs.py

Claim boundary

This is a doctrine and boundary-test tranche. It does not add live NSR execution, model calls, network calls, tool calls, schema promotion, ontology promotion, or policy admission.

Expected validation

• python3 -m pytest tests/test_neurosymbolic_boundary_docs.py
• existing repo checks

Acceptance criteria

• Superconscious remains read / reflect / propose.
• NSR methods are classified by role, not authority.
• Negative rules block soft-score truth, neural-output evidence, learned-rule schema, symbolic-derivation admission, ungrounded symbol promotion, and symbolic-policy live control.
• Example task records SocioSphere PR #388 as upstream CHRONOS capture without treating SocioSphere as schema authority.