Clarify final UI redesign sequencing

Co-authored-by: Copilot <[email protected]>

PLAN.md

@@ -907,11 +907,19 @@ SSID/password changes likely stay out of scope (higher-risk to expose over a UI
 the Pi out of its own network if fat-fingered) unless it turns out to be easy/safe to add later.
 
 ### Phase 22: Finish web UI redesign — 🚧 PLANNED
+**Roadmap decision 2026-08-09: keep this as Phase 22 rather than moving it forward.** Phase 15
+already pulled the architecture-level work forward: every page now shares one Jinja shell and
+the later list-based features have a reusable component contract. Phases 16-21 still add most of
+the final screens and dashboard data. Redesigning their visual details before those features
+exist would mean guessing at layouts and polishing the same surfaces twice; keeping the final
+pass here lets it cover the complete product once, while new phases still build on consistent
+foundations in the meantime.
+
 The current `templates/`/`static/` UI is an old, mostly-unmodified admin dashboard template
 (SB Admin 2-style Bootstrap theme) with leftover placeholder cruft never replaced with real
 content - e.g. `index.html`'s "Placeholder button", dropdown menus full of "Action" / "Another
-action" / "Something else here" stub links, and a `/dtc` page that's just an empty "Output" card
-with nothing wired up to it at all. Phase 15 already pulled the *foundational* piece of this
+action" / "Something else here" stub links, and a `/dtc` page whose new shared empty-state table
+still has no live DTC data wired into it. Phase 15 already pulled the *foundational* piece of this
 work forward (shared layout/list component, used by every phase since); this phase is the
 remaining cleanup once all the functional phases above have landed:
 - Replace/remove all remaining placeholder buttons, dropdown menus, and stub links that don't do
@@ -922,8 +930,10 @@ remaining cleanup once all the functional phases above have landed:
 - Final visual polish pass across every page now that the full feature set (dashcam/video
   browsers, trip history, settings, alerts) exists, rather than redesigning against a moving
   target.
-- Not yet scoped: whether to keep Bootstrap 5 + the current theme framework and just clean it up,
-  or replace the frontend stack entirely - decide once everything above has landed.
+- **Frontend stack decision**: keep Bootstrap 5, Jinja, and the Phase 15 shared shell/components;
+  clean up or replace the remaining SB Admin theme styling within that stack. A wholesale
+  framework rewrite would add deployment weight and offline assets on the Pi without improving
+  the tracker features, and would discard the foundation deliberately built before Phases 16-21.
 
 ### Phase 23: Native Android companion app — 🚧 PLANNED (low priority, do last)
 Revives the phone-relay goal (Phase 11/12) properly this time, using a real native app instead of