feat(agents): resolve five open Scenario D UX decision points for orientation-first review loop

Tracks WI-01 from the orientation-first review integration epic (#2090).

Five open UX decision points surfaced in the explainer/walk-back subagent research need to be finalized before the Scenario D loop can be considered complete. These decisions govern how the human-steered bookmark → dispatch → walk-back loop behaves in edge cases and how the orchestrator communicates state to the user.

**Acceptance criteria:**
- All five open Scenario D UX decision points identified in `.copilot-tracking/research/2026-06-19/pr-walkthrough-into-code-review-agent-research.md` are explicitly resolved with documented choices.
- Decisions are reflected in the relevant agent files (`code-review.agent.md`, `code-review-explainer.agent.md`, `code-review-walkback.agent.md`) as behavioral constraints or instructions.
- No contradictory interaction patterns remain between the orchestrator and its subagents.

**References:**
- Parent epic: #2090
- Research: `.copilot-tracking/research/2026-06-19/pr-walkthrough-into-code-review-agent-research.md`
- Backlog research: `.copilot-tracking/research/subagents/2026-06-20/current-review-state-and-backlog-research.md`
Related to #2090

> Generated by [Issue Triage](https://github.com/microsoft/hve-core/actions/runs/27888664228) for issue #2090 · 70.8 AIC · ⌖ 13.1 AIC · ⊞ 28.5K · [◷](https://github.com/search?q=repo%3Amicrosoft%2Fhve-core+is%3Aissue+%22gh-aw-workflow-call-id%3A+microsoft%2Fhve-core%2Fissue-triage%22&type=issues)

Start by reading .copilot-tracking/research/2026-06-19/pr-walkthrough-into-code-review-agent-research.md and the backlog research to identify the five Scenario D UX decision points. Then inspect code-review.agent.md, code-review-explainer.agent.md, and code-review-walkback.agent.md. Done means all five choices are documented in the relevant agent files with no contradictory interaction patterns.

Written by the indexing model from the issue text.