Summary

• Bump the shared review result contract to schema 1.2 (additive only): add
  a consumer_impact_evidence array recording, per changed shared symbol
  whose other call sites/consumers were traversed, the symbol, its defining
  location, one or more concrete consumer_search_evidence entries, and a
  disposition of all_consumers_consistent, inconsistency_found, or
  no_other_consumers.
• review-suite/scripts/validate.py: enforce structure and non-emptiness
  only. Only correctness or aggregate results may carry this evidence. A
  disposition claiming other consumers were found
  (all_consumers_consistent, inconsistency_found) requires search
  evidence covering more than the changed symbol's own location;
  no_other_consumers requires at least one concrete search. The validator
  deliberately does not determine which changed symbols require an entry —
  it receives only a packet and a result, with no repository checkout to
  search, so it cannot itself decide whether a changed symbol has other call
  sites; that judgment belongs to the lens performing the traversal (Add correctness traversal and verification-sufficiency passes #53,
  not yet implemented). Extend the stale-schema-version rejection so a v1.1
  result now fails with a useful migration error, mirroring the existing
  v1.0 rejection.
• Add a new canonical fixture, consumer-impact-traversal, modeling a
  sanitized version of the baseline's dependency-strictness-propagation
  miss, covering all 5 required fixture/test scenarios in
  review-suite/scripts/tests/test_contracts.py.
• Migrate every canonical fixture, eval helper, and each bundled skill's
  standalone eval fixtures to schema 1.2; refresh the four bundled
  review-suite copies via just sync-contracts.
• Document the validator's scope boundary explicitly in CONTRACT.md, the
  validator docstring, and a dedicated regression test, after a first
  review-code-change pass raised — and a second, clean pass confirmed
  resolved — the question of whether the validator should itself detect
  missing traversal from the diff (it should not: that would be exactly the
  independent correctness explorer / static call-graph tooling this
  ticket's non-goals rule out).

Why

• Add consumer/impact-traversal evidence to the shared review contract #52 (narrowed by Use the v1 baseline to finalize and preregister the review v2 graph #59's decision record): the frozen v1 baseline's
  dependency-strictness-propagation case shows 5 of 5 attempts missed the
  same sibling call site of a changed shared helper, because nothing in the
  contract required a reviewer to record whether other call sites were
  traversed. This ticket owns only the schema and validator for that
  evidence; it does not add the lens behavior that populates it (Add correctness traversal and verification-sufficiency passes #53), and
  it does not add the changed-surface ledger, acceptance trace, or risk
  profile Use the v1 baseline to finalize and preregister the review v2 graph #59 explicitly dropped as unsupported by baseline evidence.

Acceptance criteria (from #52's body)

• Schema version 1.2 adds exactly the consumer_impact_evidence array
  described in the ticket; no changed-surface ledger, acceptance trace,
  or risk profile is added.
• The validator requires a non-empty, structurally valid entry, and
  accepts a concretely-evidenced no_other_consumers disposition
  otherwise (structure and non-emptiness only — see the CONTRACT.md
  "Consumer/impact evidence" section for the validator's scope boundary
  and why determining which symbols require an entry is lens
  judgment, not validator judgment).
• All required positive and negative fixtures exist (5 scenarios,
  covered by ConsumerImpactEvidenceTests in test_contracts.py).
• Canonical and bundled contracts are byte-identical after
  just sync-contracts.
• Schema 1.1 fixtures and callers are migrated atomically; stale
  documents (both 1.0 and now 1.1) fail with a useful error.
• just format, just lint, and just test pass.
• README and CHANGELOG describe the evidence addition and version
  transition (CHANGELOG updated; root README does not document schema
  versions, matching Make clean verdicts require passing validation and current-head lens evidence #51's own precedent).
• No lens prompt, rubric, or orchestration behavior changes in this
  ticket.

Test plan

• just format — all checks passed, no reformatting needed
• just lint — all checks passed (skills-ref validation of all 7
  skills, plugin packaging validation)
• just test — 234 tests in review-suite/scripts/tests (up from 227),
  plus every skill's own tests, all OK
• just sync-contracts — all four bundled review-suite copies
  byte-identical to canonical source (verified by
  test_bundled_contracts.py and a direct cmp check)
• Reviewed by the repository's own review-code-change suite (2 rounds;
  round 1 raised one correctness finding proposing the validator detect
  missing traversal from the diff — declined with evidence as exactly
  the independent-explorer/static-call-graph machinery this ticket's
  non-goals forbid, and strengthened documentation instead; round 2
  clean across solution-simplicity, correctness, and code-simplicity)

Scope note

This PR does not touch review-suite/evals/baseline/v1/ or
review-suite/evals/v2/ (frozen artifacts from #58/#59), does not touch
review-code-change/SKILL.md's orchestration matrix, the correctness
rubric, or any lens prompt, and does not add lens behavior that populates
consumer_impact_evidence — that is #53's scope.

Fixes #52