feat: migrate on-device Qwen3 ASR to CoreML for background Flow dictation

Replace MLX GPU inference with CoreML bundles so transcription continues
while the host app is backgrounded. Adds model download and warm-up,
vendored Qwen3Speech, and updates onboarding, settings, and copy for the
~1.6 GB CoreML package (iOS 18+).

OSGKeyboard/Views/PrivacyPolicyView.swift

@@ -0,0 +1,33 @@
+// PrivacyPolicyView.swift
+// OSGKeyboard · Main App
+//
+// Bundled privacy policy HTML. Reached from Settings → About.
+
+import SwiftUI
+import OSGKeyboardShared
+
+struct PrivacyPolicyView: View {
+    @Environment(\.themePalette) private var palette: ThemePalette
+    @ObservedObject private var config = ProviderConfig.shared
+
+    var body: some View {
+        LegalWebView(
+            resourceName: "PrivacyPolicy",
+            scrollToAnchor: privacyScrollAnchor
+        )
+        .background(palette.background.ignoresSafeArea())
+        .navigationTitle("settings.privacy.policy")
+        .navigationBarTitleDisplayMode(.inline)
+    }
+
+    private var privacyScrollAnchor: String? {
+        switch config.uiLanguage {
+        case .chinese:
+            return "zh"
+        case .english:
+            return "top"
+        case .auto:
+            return config.uiLanguage.resolvedLanguageCode().hasPrefix("zh") ? "zh" : "top"
+        }
+    }
+}