Migrate RFC review cards to planning metadata

[SoundBlaster]

Motivation

PR #9 establishes the planning metadata contract, but the canonical 60-card backlog still needs to adopt it before dependency-aware planning can be trusted.

Goals

• Migrate every review card to explicit profile, dependency, maturity, release, and evidence metadata.
• Derive reverse dependencies and a deterministic ready/blocked queue without persisting redundant state.
• Make the canonical repository gate reject any fallback to transitional metadata.

Changes

• Assign all 60 cards to 10 primary profiles with 113 cycle-free hard dependency edges.
• Initialize 30 present cards as specified and 30 partial/missing cards as proposal; leave all target releases unassigned.
• Attach 178 existing RFC anchor references as evidence for present/partial coverage.
• Derive blocks and readiness during the build; 44 cards are ready and 16 are blocked by incomplete prerequisites.
• Document the source-versus-derived planning model and enforce canonical planning_metadata_mode: required while retaining schema-v2 transition compatibility.

Validation

• make review-check PYTHON=/tmp/agent-surface-maintenance-uv/bin/python (29 tests)
• /tmp/agent-surface-maintenance-uv/bin/python -m json.tool review/review-data.json
• /tmp/agent-surface-maintenance-uv/bin/python -m json.tool review/review-data.schema.json
• git diff --check
• Independent semantic review of hard dependencies and readiness: no remaining P0–P2 findings.

Boundaries and Non-Goals

• Does not change RFC prose, card priorities, or existing coverage statuses.
• Does not assign cards to release 0.2; target_release remains null until sequencing is decided.
• Does not persist blocks or readiness and does not infer maturity from generated artifacts.
• Does not add dashboard filters or metadata UI; that is the next stacked PR.

Notes

• Stacked on PR Add RFC review metadata validation #9 (codex/review-metadata-schema).
• Readiness depends only on direct prerequisites being present and at least specified; a backlog card's own status does not prevent it from being ready to work on.