#11734: CLI: scope /model and /new prompt messaging to fallback-only paths

## What / Why This PR tightens two UX paths so model messaging only changes when fallback behavior is actually active, while preserving existing behavior on the normal path. ### What changed 1. **Fallback-scoped `/model` messaging** - Adjusts `/model`/status messaging so fallback-related wording is shown only when fallback is in use. - Non-fallback `/model` behavior remains aligned with `main`. 2. **`/new` prompt model line cleanup** - Updates reset prompt wording to only mention model differences when `current model != configured default`. - If current and default match, no extra model line is shown. ### Why - Prevents confusing model/fallback language in non-fallback scenarios. - Keeps user-facing messaging accurate to runtime state. - Limits this change to a focused UX correctness fix without broad behavior changes. ## Focused scope statement This PR is intentionally limited to: - fallback-only model/status messaging behavior, and - `/new` prompt wording for model-difference clarity. No broader model selection flow, architecture, or unrelated UX changes are included. ## Local validation Executed locally for this branch: ```bash pnpm build && pnpm check && pnpm test ``` - `pnpm build` ✅ - `pnpm check` ✅ - `pnpm test` ✅ ### Manual local-instance validation Validated behavior in a local OpenClaw instance: - Confirmed fallback-only wording appears only when fallback is active. - Confirmed non-fallback paths match `main` behavior. - Confirmed `/new` model line only appears when current/default differ. ## AI-assisted disclosure - [x] AI-assisted PR (description + implementation support) - [x] Testing level: **fully tested locally** (`pnpm build && pnpm check && pnpm test` + manual local instance validation) - [ ] Prompts/session logs attached (not included in this PR) - [x] Author reviewed and understands the code and behavior changes