fix(chargen): Campaign CC CC6b-MOUNT re-review residuals R1-R3 + nits — REVIEW-CLOSED

R1: LivePresentationComposition's chargen-preview diagnostic fired on every
ordinary launch without ACDREAM_RETAIL_UI set, since interaction.RetainedUi
is null in that configuration and there's no Appearance page to warn about.
Narrowed the else-if guard to require RetainedUi is not null too, so the
diagnostic only fires in the one configuration it actually diagnoses.

R2: filed AP-221 for the F8 one-shot-binding disposition the re-reviewer
accepted as scoped but which shipped without its own register row — the
chargen preview's GPU-side binding reads the retryable mount coordinator's
widget exactly once, so a slow-DAT frame permanently kills the preview for
the session with only R1's diagnostic as evidence.

R3: rewrote AP-217 after re-deriving from the decomp. The original row
claimed the GradCircle was an interactive click-to-hue picker with no
handler wired up. gmCGAppearancePage::ListenToElementMessage's dispatch
switch has no case for the GradCircle's offset at all — it isn't a click
target in retail either. DoGradDisk is a paint-only routine that blits the
gradient art tinted with the current color (or blanks it for Eyes)
whenever SetColor/SetSelection run. acdream's real gap is that it never
repaints the GradCircle — a cosmetic paint gap, not a dead control.

N1: tightened AP-220's "leaving Gearknight for something else" — the
decomp shows leaving Gearknight for Olthoi/OlthoiAcid takes a separate
branch that does not randomize; only leaving for a non-Olthoi heritage
does.

N2: added the requested media pin to the F2 spin-highlight live-DAT test,
then measured it against the installed EoR dat rather than assuming it
would pass. It doesn't: none of the nine spins author Highlight-state
media on either consumed arrow face segment, so
TrySetRetailState(Highlight) is a silent no-op for all of them today.
Pinned the test to the measured reality (ActiveState stays "Normal") and
filed AP-222 documenting the discovery — unresolved whether retail's own
spin art has the same gap.

Plan doc: CC6b-MOUNT ledger row updated to REVIEW-CLOSED with the full
commit chain and re-review disposition; OWED list corrected for AP-217/
AP-222.

Gates: dotnet build -c Release green. App suite 5223 passed / 3 skipped
(ACDREAM_PROBE_LIVE_MOUNT=1, ACDREAM_DAT_DIR set) — count held exactly at
baseline. Runtime suite 1713/0 — unchanged.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

src/AcDream.App/Composition/LivePresentationComposition.cs

@@ -1078,13 +1078,25 @@ internal sealed class LivePresentationCompositionPhase
                     }
                 });
         }
-        else if (dispatcherLease.Resource is not null)
+        else if (dispatcherLease.Resource is not null && interaction.RetainedUi is not null)
         {
             // Fix round F8: dispatcher is available but the mount coordinator
             // hadn't resolved ChargenPreviewViewportWidget by this one-shot
             // pass — loud instead of silent, since the coordinator's own
             // later per-frame retries cannot recover this GPU-side binding
             // (see this block's own disposition comment above).
+            //
+            // Re-review R1: the retained UI arm (`interaction.RetainedUi`)
+            // is null in the default configuration (ACDREAM_RETAIL_UI
+            // unset — see InteractionRetainedUiComposition.cs's own gate on
+            // RuntimeOptions.RetailUi), and in that configuration there is
+            // no Appearance page at all. The dispatcher lease is
+            // acquired unconditionally regardless of retained-UI presence,
+            // so without this second guard every ordinary launch printed
+            // this diagnostic even though nothing was actually broken.
+            // Narrowed to fire only in the one configuration it is meant to
+            // diagnose: retained UI mounted, dispatcher ready, but the
+            // coordinator's widget resolution missed this one-shot pass.
             Console.WriteLine(
                 "[UI] chargen preview viewport unavailable at composition "
                 + "time — the Appearance page's zoom/rotate controls and "

tests/AcDream.App.Tests/UI/Layout/CharacterCreationLiveDatTests.cs

@@ -370,6 +370,36 @@ public sealed class CharacterCreationLiveDatTests
             // drops it shows up here instead of as a silently-dead
             // highlight.
             Assert.True(spin.ToggleBehavior, $"spin 0x{spinId:X8} must author ToggleBehavior for the current-part highlight to work.");
+            // Re-review nit N2 (2026-08-15): ToggleBehavior alone is necessary
+            // but not sufficient — UiButton.UpdateVisualState only COMMITS the
+            // requested state when _availableStates actually contains it
+            // (UiButton.cs's TrySetRetailState -> Selected setter ->
+            // UpdateVisualState chain). MEASURED (not assumed) against the
+            // installed EoR dat: TrySetRetailState(Highlight) itself always
+            // reports success (the ToggleBehavior branch commits
+            // unconditionally, matching TrySetRetailState's own contract),
+            // but NONE of the nine spins actually carries Highlight /
+            // Highlight_rollover / Highlight_pressed media on either of
+            // their two consumed arrow face segments — every one of them
+            // authors only Normal / Normal_rollover / Ghosted. So F2 item 2's
+            // current-part highlight is CURRENTLY A NO-OP for every spin:
+            // ActiveState silently stays at its prior value ("Normal")
+            // instead of ever becoming "Highlight". The pre-existing
+            // ToggleBehavior pin above never caught this because it only
+            // checks the PROPERTY that gates the state-machine branch, not
+            // whether that branch has anything to actually draw. Filed as
+            // AP-222 (retail-vs-acdream status unresolved — retail's own
+            // gmCGAppearancePage::SetSelection call sites are cited for
+            // the SetState(1)/SetState(6) calls, not for whether retail's
+            // OWN spin art authors Highlight media either). Pinned to
+            // "Normal" so a future DAT revision that adds real Highlight
+            // media is what makes this assertion start failing — the
+            // correct trigger to update it to "Highlight" instead of a
+            // silently-reintroduced dead highlight going unnoticed either way.
+            Assert.True(
+                spin.TrySetRetailState(UiButtonStateMachine.Highlight),
+                $"spin 0x{spinId:X8} must accept a Highlight state request.");
+            Assert.Equal("Normal", spin.ActiveState);
         }
 
         // Every color-wheel-family id resolves through EXISTING