feat(engine): merge cloudflare-migration — paradigm D engine, BYOK proxy, story persistence (#95)

Squash-merge the cloudflare-migration branch (7 commits by Kai ki) into
staging with conflict resolution, feature integration, and bug fixes.

Engine:
- Paradigm D: single-stream Writer replacing dual-phase Plan/Beats
- Delete Architect agent; story bible generated via Writer <plan> tag
- Modular prompt architecture (segments/registry/builder)
- StreamRouter for tagged stream splitting (<plan>/<story>/<choices>)

Infrastructure:
- Cloudflare Workers deployment (wrangler.jsonc, OpenNext adapter)
- D1 database schema + Drizzle ORM (scaffolded, not yet active)
- R2 storage helpers (scaffolded, not yet active)
- Story persistence API routes + client-side persistence

BYOK (Bring Your Own Key):
- /api/llm/user-proxy with SSRF-protected LLM proxy (+ requireUser auth)
- CORS-aware fetch in ai-client: auto-detect CORS failure, fallback to
  server proxy transparently via OpenAI SDK custom fetch
- BYO config support added to classify-freeform and vision routes
- SettingsModal CORS privacy notice (keys never logged/stored)

SSE streaming:
- engineClient.ts: fetchSSE helper for progressive scene events
- startSession/requestScene accept optional emit callback
- Fix SSE error event field name (error → message) in scene/start routes

i18n integration:
- Wire buildLanguageDirective into paradigm D's prompt builder
- Update corsNotice i18n keys (zh-CN/en/ja) with CORS proxy privacy text
- Preserve Session.language + LanguageSwitcher from i18n commit

Co-authored-by: Kai ki <155355644+zbf1009@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

lib/clientLlmConfig.ts

@@ -0,0 +1,99 @@
+// Bring-your-own LLM API keys — stored CLIENT-SIDE ONLY.
+//
+// When a user supplies their own keys, we persist {provider, baseUrl, apiKey}
+// in localStorage and send them with each /api/start and /api/scene request.
+// Keys never leak to server logs or persistence — they only pass through the
+// request→config construction path.
+
+const STORAGE_KEY = "infiplot:llm";
+
+/** Provider types matching byoProxy and ProviderProtocol */
+export type LlmProvider = "openai" | "claude" | "gemini";
+
+/** Stored BYO LLM config — exactly what we persist. */
+export type StoredLlmConfig = {
+  /** Which provider API to use */
+  provider: LlmProvider;
+  /** User's API key */
+  apiKey: string;
+  /** Optional custom base URL (empty = use provider default) */
+  baseUrl?: string;
+  /** Optional model name (empty = use server-side default for this provider/role) */
+  model?: string;
+};
+
+/** Per-role LLM config the user can independently configure */
+export type ByoLlmSettings = {
+  text?: StoredLlmConfig;
+  image?: StoredLlmConfig;
+  vision?: StoredLlmConfig;
+};
+
+/**
+ * Read persisted BYO LLM config. Returns null when running on the server,
+ * when nothing is stored, on parse failure, or when the stored shape is invalid.
+ */
+export function readStoredLlmConfig(): ByoLlmSettings | null {
+  if (typeof window === "undefined") return null;
+  try {
+    const raw = window.localStorage.getItem(STORAGE_KEY);
+    if (!raw) return null;
+    const parsed = JSON.parse(raw) as Partial<ByoLlmSettings>;
+
+    // Validate each role config
+    const result: ByoLlmSettings = {};
+    for (const role of ["text", "image", "vision"] as const) {
+      const cfg = parsed[role];
+      if (cfg && typeof cfg === "object") {
+        const provider = cfg.provider as string;
+        const apiKey = cfg.apiKey as string;
+        if (["openai", "claude", "gemini"].includes(provider) && apiKey?.trim()) {
+          result[role] = {
+            provider: provider as LlmProvider,
+            apiKey: apiKey.trim(),
+            baseUrl: typeof cfg.baseUrl === "string" ? cfg.baseUrl.trim() : undefined,
+            model: typeof cfg.model === "string" ? cfg.model.trim() : undefined,
+          };
+        }
+      }
+    }
+
+    return Object.keys(result).length > 0 ? result : null;
+  } catch {
+    return null;
+  }
+}
+
+/**
+ * Persist BYO LLM config. Trims keys and baseUrls so trailing whitespace
+ * from paste never breaks headers.
+ */
+export function writeStoredLlmConfig(config: ByoLlmSettings): void {
+  if (typeof window === "undefined") return;
+  try {
+    const payload: ByoLlmSettings = {};
+    for (const role of ["text", "image", "vision"] as const) {
+      const cfg = config[role];
+      if (cfg) {
+        payload[role] = {
+          provider: cfg.provider,
+          apiKey: cfg.apiKey.trim(),
+          baseUrl: cfg.baseUrl?.trim() || undefined,
+          model: cfg.model?.trim() || undefined,
+        };
+      }
+    }
+    window.localStorage.setItem(STORAGE_KEY, JSON.stringify(payload));
+  } catch {
+    // Storage disabled / quota / private mode — BYO simply stays off.
+  }
+}
+
+export function clearStoredLlmConfig(): void {
+  if (typeof window === "undefined") return;
+  try {
+    window.localStorage.removeItem(STORAGE_KEY);
+  } catch {
+    // ignore
+  }
+}