feat(code-review): render a two-part Technical + Functional review on every CR run

[pekral]

Summary

Closes #737.

Every full-lens CR wrapper (code-review, code-review-github, code-review-jira, code-review-bugsnag) now produces an explicit two-part output on every run:

1. Technical review — the existing strict rule-compliance analysis, unchanged in behavior. Published on the PR comment.
2. Functional review — a full acceptance-criteria checklist (Met / Not met / Partial / Divergent per criterion) plus an explicit Goal met: Yes/No verdict. Built by assignment-compliance-check, published on the linked-tracker comment via pr-summary, with the one-line verdict mirrored onto the PR comment's Summary line.

Key behavior change: assignment-compliance-check now renders the block on every run that has a linked tracker — including the fully affirmative report on a clean run — instead of the old "gaps-only, omit when clean" behavior. This is a deliberate, narrow exception to the project's "report only what needs action; never render a positive banner; omit when clean" convention, scoped to the Functional-review verdict alone; every other CR section (Technical review, Architecture, Coverage, …) keeps that convention unchanged. pr-summary's {assignment_verdict} top banner gained the matching affirmative variant (English + Czech).

api-review is a documented carve-out (per orchestrator decision): instead of invoking the full assignment-compliance-check, it renders a light API contract matches assignment: Yes/No verdict derived from its own Core Checks walk, so API-focused issues get a functional signal without running two full assignment-conformance engines over the same diff.

Files changed

• rules/code-review/general.mdc — canonical Two-part CR output — Technical & Functional review contract (replaces the old Assignment Conformance section).
• skills/code-review/SKILL.md — thin reference to the contract (stays under the 5000-token skill-check budget; verified str_word_count < 5000 in a pinning test).
• skills/assignment-compliance-check/SKILL.md — core behavior change: full checklist + always-render + affirmative verdict.
• skills/pr-summary/SKILL.md + templates/pr-summary-{github,jira}.md — affirmative {assignment_verdict} banner variant.
• skills/code-review-{github,jira,bugsnag}/SKILL.md — updated wrapper prose describing the new always-render behavior.
• 4 wrapper output templates (code-review, code-review-github, code-review-jira, code-review-bugsnag) — framed as "Technical review" with a pointer to the Functional review.
• skills/api-review/SKILL.md + templates/review-output.md — light functional cross-check carve-out.
• CHANGELOG.md — feat(code-review) entry.
• tests/Installer/CodeReviewContentTest.php, tests/Installer/SkillsContentTest.php — updated pins for the reversed convention (old "omit when clean" pins replaced) + new pins for the two-part contract, per the project's content-pin-tests-are-the-contract-enforcement convention.

How to test

1. Run composer build — skill-check (0 errors/warnings, 62 skills), phpcs, Pint, Rector, phpstan, security-audit, and the full Pest suite (329 tests, 100% coverage) all pass.
2. Read rules/code-review/general.mdc Two-part CR output — Technical & Functional review and confirm it defines both halves, the affirmative exception, and the api-review carve-out.
3. Trigger a CR run on a PR linked to a GitHub issue (or JIRA ticket) and confirm: the PR comment is framed as the Technical review; the linked-tracker comment carries a full Goal met: Yes/No checklist (including Met entries) via pr-summary; a clean run renders ✅ Changes satisfy the assignment — all N acceptance criteria met instead of nothing.

TODO

• code-review-bugsnag's PR-comment summary line does not carry an assignment conformance: … mirror slot the way the other three wrappers do (pre-existing asymmetry, unrelated to this change) — worth a follow-up if full parity across all four wrappers' summary lines is desired.

[pekral]

Code Review

Technical review. This comment is the Technical review half of the two-part CR output (@rules/code-review/general.mdc Two-part CR output — Technical & Functional review) — strict rule compliance only. The Functional review (full acceptance-criteria checklist, Goal met: Yes/No) is published separately on the linked-issue comment via @skills/pr-summary/SKILL.md.

Status: Converged — 0 Critical / 0 Moderate (iteration 2, commit 42f8b7a).

All four iteration-1 findings are resolved and no regressions were introduced:

• M1 (api-review template pin) — the two-part test now reads skills/api-review/templates/review-output.md and pins **API contract matches assignment:**.
• M2 (code-review-bugsnag drift) — full parity: the bugsnag template's Summary line and header-note now carry the same assignment conformance: mirror as the other three wrappers, and the pin asserts the slot across all four templates.
• Minor 1 (api-review light verdict render target) — a Render target — standalone runs only clause was added to the rule, the skill, and the template so the light verdict self-suppresses on an inline sub-lens invocation; pinned.
• Minor 2 (Critical-fold clause) — pinned in rules/code-review/general.mdc.

Consolidated security review (athena, commit 42f8b7a): 0 Critical / 0 Moderate / 0 Minor — documentation / CR-infra prose only, no new sink.

---

Summary: 0 Critical · 0 Moderate · 0 Minor · 0 Refactoring · assignment conformance: conformant