Add Human Protection Layer checks to cognition loop

[mdheller]

Purpose

Turn docs/human-protection-layer-adoption.md into Superconscious planning and trace checks.

Scope

• Add HPL scope assessment before protected-person or human-impacting tool requests.
• Add blocked status for missing policy admission.
• Ensure planning cannot authorize execution.
• Ensure private evidence is not emitted in safe operational traces.
• Ensure unsupported mechanism labels cannot become planning assumptions.
• Ensure memory write proposals require explicit memory decision.
• Ensure HPL status appears in replay and benchmark artifacts.

Acceptance criteria

• HPL adoption doc is linked from docs index or README.
• Planning trace includes HPLScope.assessed.
• Side-effectful tool requests require policy admission in fixtures.
• Human actuation plans are blocked or marked external-review-required.
• Private evidence does not appear in safe traces.
• Benchmark fixture asserts blocked status is preserved.

Non-goals

• No authorization authority added to Superconscious.
• No schema ownership moved into Superconscious.
• No network/model/host side effects added to M1.