chore(ui): remove dead banner ViewBuilder params from the Daily hubs

[wei18]

Closes #1096
part of #1014

What

Drop the Banner generic and the banner: ViewBuilder param from DailyHubShellView (GameShellKit), DailyHubView (SudokuKit) and MinesweeperDailyHubView (MinesweeperKit), plus the two forwarding call sites. No production caller has passed one since #1020; the banner lives in the shared tabViewBottomAccessory. Today-case comments in both Live+TabRoots.swift are rewritten as mirrors and no longer describe a dead param. GameShellKit stays zero-dependency.

Kept on purpose: the now single-child VStack(spacing: 0) in DailyHubShellView.body, so the structure TXT moves on line 1 only (PM-accepted).

Finding: one test DID pass a banner

BannerSlotDarkBandRegressionTests (the #851 dark-band pin) built DailyHubView(viewModel:) { BannerSlotView(…) } through a trailing closure, so the issue's banner: regex could never match it. PM ruling (A): keep the pin, reconstruct the pre-#1024 in-shell composition in the fixture (VStack(spacing: 0) { DailyHubView; BannerSlotView }.background(page token)). Proof it is faithful: base-rendered vs head-rendered PNG for postFix_themedBackground_dailyHubDarkMode_bannerOpen_noBand are cmp byte-identical.

Other test touches: DailyHubViewTabReturnRetryTests.swift DailyHubView<EmptyView> → DailyHubView (compile fix). DailyHubShellViewGenericityTests needed no edit — its 3 sites infer the generics.

Snapshot changes — 29 .txt, line 1 only, 0 PNG

git diff --stat origin/main -- '**/__Snapshots__/**' → 29 files, 1 insertion + 1 deletion each, all .txt.

Files	Old line 1 (pattern)	New line 1 (pattern)
SudokuKit DailyHubViewTests (10)	…GV8SudokuUI12DailyHubViewVS_9EmptyView_GVS_30_EnvironmentKeyWritingModifier…GS5_…	…V8SudokuUI12DailyHubViewGVS_30_EnvironmentKeyWritingModifier…GS4_…
SudokuKit TodayTabViewTests (5)	…TodayTabHostO8SudokuUI8AppRouteGVS4_12DailyHubViewVS_9EmptyView__GVS_30_…GS8_…	…TodayTabHostO8SudokuUI8AppRouteVS4_12DailyHubView_GVS_30_…GS7_…
MinesweeperKit MinesweeperDailyHubSnapshotTests (9)	…NavigationPathGV13MinesweeperUI23MinesweeperDailyHubViewVS_9EmptyView__GVS_30_…GS7_…	…NavigationPathV13MinesweeperUI23MinesweeperDailyHubView_GVS_30_…GS6_…
MinesweeperKit MinesweeperTodayTabViewTests (5)	…TodayTabHostO13MinesweeperUI8AppRouteGVS4_23MinesweeperDailyHubViewVS_9EmptyView__…	…TodayTabHostO13MinesweeperUI8AppRouteVS4_23MinesweeperDailyHubView_…

The root type loses its EmptyView generic argument; the mangling substitution indices after it renumber (e.g. GS5_→GS4_, GS8_→GS7_, GS7_→GS6_); exact prefixes vary per fixture wrapper. Store-frame and ASC-emit suites are green and their baselines untouched.

Spliced, not re-recorded. This host is on macOS 27.0 (#1097): all 29 files are already red on clean origin/main from renderer drift (line 4 NSClipView → SwiftUI.HostingClipView, or a 1pt _FocusRingView y-shift). A record-mode run would have baked that drift into 29 baselines. Instead each file = head-rendered line 1 + committed lines 2+. Proofs:

1. Line 1 is OS-independent: base-rendered line 1 (macOS 27) == committed line 1 on origin/main, 29/29.
2. For all 29 files, diff(spliced, head-rendered) == diff(origin/main committed, base-rendered) — drift lines only, no line 1.

Base vs head (no-record), same commands and exact filters, sequential

Base = clean detached tree at 331b4de7. Post = this branch after the splice.

Suite	BASE	HEAD (post-splice)
GameShellKit full	124/124 pass	124/124 pass
MinesweeperKit full (412 tests)	failed, 59 issues	failed, 59 issues — identical failing-test set; 44/44 structure-diff lines identical
SudokuKit --filter DailyHub (76)	failed, 20 issues	failed, 20 issues — identical set; 22/22 identical
SudokuKit --filter TodayTab (5)	failed, 10 issues	failed, 10 issues — identical set; 10/10 identical
SudokuKit --filter ASCScreenshotEmit (9)	pass	pass
SudokuKit --filter BoardLoaderViewDailyPrecheckTests (6)	pass	pass
SudokuKit --filter BannerSlotDarkBandRegressionTests (1)	failed, 1 issue (PNG)	failed, 1 issue (PNG)

These runs are red, and red identically on base — they are not green.

#1097 (environment) — base-identical failures, not from this PR

• Affected suites: 30 PNG (DailyHubViewTests 10, TodayTabViewTests 5, MinesweeperDailyHubSnapshotTests 9, MinesweeperTodayTabViewTests 5, BannerSlotDarkBandRegressionTests 1) + the drift lines in the same 29 TXT.
• Untouched MinesweeperKit suites red on both: ProgressScreenTests, MinesweeperCompletionSnapshotTests, MinesweeperPracticeHubSnapshotTests, MinesweeperControlClusterSnapshotTests, MinesweeperBoardLoaderViewFailedExitTests.
• Rendered-vs-rendered: the failing output each run produced was copied out of swift-snapshot-testing's shared temp dir (head before the base run overwrote it). 30/30 PNG cmp byte-identical base↔head; 29/29 TXT differ at exactly 1c1. Zero PNG failures exist only at head. No PNG was re-recorded.

Positive control for the banner: search

git grep -n -E '(^|[^A-Za-z_])banner:' -- '*.swift'

• origin/main: hits in DailyHubShellView.swift (:65, :94), DailyHubView.swift (:66, :80, :307), MinesweeperDailyHubView.swift (:48, :58, :174) and the Today-case comments in both Live+TabRoots.swift.
• head: no hit in the three types, their callers' Today case, or the touched tests. Remaining hits in the two Live+TabRoots.swift / LiveRouteFactory.swift files are the Practice/Settings history comments from feat(monetization): #1024 follow-through after #1062 merges — merge main, simulator evidence, open PR #1080/feat(monetization): move banner to tabViewBottomAccessory, with tab-bottom fallback #1024, out of scope here.
• The regex is blind to trailing-closure callers (see Finding above); the compiler is the real proof — all three packages build with the param gone.

Other gates

• swiftlint --strict on changed Swift files: 0 violations. scan:hygiene, gitleaks: pass.
• iOS Simulator builds (Debug, generic destination): Sudoku ** BUILD SUCCEEDED **, Minesweeper ** BUILD SUCCEEDED ** (after tuist generate).

🤖 Generated with Claude Code

[wei18]

PM review record, head 2326b962: CR + acceptance ACCEPT / MERGE.

• Compile proof, done because a banner: regex cannot see trailing-closure callers: swift build --build-tests succeeds for GameShellKit, SudokuKit and MinesweeperKit. None of the three Daily types declares Banner any more. All 29 DailyHub(Shell)?View construction sites were read, and none passes a banner.
• No production behaviour change. Non-snapshot hunks are the param and generic removal, the forwarding call-site removal, comments, and test fixtures only. DailyHubShellView.body keeps its single-child wrapper, as ruled.
• Line-1 splice: 29 .txt files, 1+/1− each, verified by script; the new line 1 drops the EmptyView generic argument with substitution indices renumbered by −1. No .png changed.
• Base vs head, reproduced independently in the reviewer's own throwaways under macOS 27: GameShellKit 124/124 on both; MinesweeperKit 59 issues with the same 37 failing names on both; SudokuKit DailyHub 20 issues (10 names), TodayTab 10 issues (5 names), dark-band 1 PNG, all identical on both. All remaining failures are test(snapshots): strict baselines fail after the macOS 27 upgrade (2026-09-16), renderer drift with no code change #1097 environment drift. On spliced files the only remaining difference is the drift line; line 1 matches on both sides.
• Dark-band fix(shell): faint dark-mode background banding on hub screens (both apps) #851 pin: the rendered PNGs from head and base are byte-identical (cmp), so the fixture reconstruction is faithful.
• Hygiene: CI 7/7, including SwiftLint (changed files) and the ASC frame-drift gate; Closes #1096 only; GameShellKit Package.swift unchanged.