Chat mode, phase 1: grounded streaming chat over your notes

Entry points (no sixth tab): Ask in the Library header (desktop button,
phone icon), Ask about this deck in the deck menu, Ask a question on
study/homework results (compact result travels as context), and a Chat
row in the desktop sidebar. Phones get a full-screen route with back +
New chat; the tab bar steps aside.

- Client-side retrieval: chunk scoped decks (~800 tokens), BM25 against
  the question + recent turns, top chunks under ~7k tokens, each
  labelled with deck + section. Notes come from the deck's stored
  sourceText (already kept for Regenerate); decks without it fall back
  to generated content and the UI says so.
- Stateless server: /api/chat validates against the CHAT_MODELS
  allowlist, trims the history to ~6k tokens, caps context at ~9k and
  streams token/thinking/done/error NDJSON. /api/chat/extract reduces
  attachments to text through the existing vision step.
- UI: user bubbles + full-width assistant text, streaming with stop,
  auto-scroll that pauses with a jump-to-bottom button, safe markdown
  (React nodes, no HTML injection; KaTeX dynamically imported only when
  a message has math), Thinking row for reasoning, copy/regenerate/
  edit-and-resend, model + context pickers (last model remembered),
  composer with visualViewport keyboard handling, safe-area, 16px
  textarea, Enter-to-send on desktop.
- Usage-limit failures keep the draft and show the shared message with
  Settings/Ollama actions; stopping maps the reader abort to a clean
  stopped state; offline disables sending with a reason.
- Settings: Keep notes text for chat toggle; deck menu: Delete notes
  text. RATE_LIMIT_CHAT + CHAT_MODELS in .env.example and README.
- Tests: chat suite (63 checks, stub Ollama) wired into run-all.
  Full set: 12/12 suites, 487 checks. SW -> freezmo-v32.

Also: tests/model-compare.mjs (diagnostic) benchmarks two models on the
same notes (latency, repair rounds, activity types) so the gpt-oss vs
gemma default can be decided on numbers; defaults unchanged.

lib/chat.ts

@@ -0,0 +1,83 @@
+"use client";
+
+/**
+ * Chat types shared by the chat UI, the context builder and the streaming
+ * client. The API key rule applies here too: nothing in a chat request may
+ * carry the key anywhere except the `x-ollama-key` header of our own route.
+ */
+
+export type ChatRole = "user" | "assistant";
+
+/** A context chunk that actually travelled with a request. */
+export type ChatSource = {
+  deckId: string;
+  deckTitle: string;
+  section: string;
+};
+
+export type ChatAttachment = {
+  name: string;
+  kind: "photo" | "document" | "text";
+  chars: number;
+};
+
+export type ChatError = { message: string; kind?: string };
+
+export type ChatMessage = {
+  id: string;
+  role: ChatRole;
+  content: string;
+  /** Reasoning streamed separately; hidden behind the Thinking row. */
+  thinking?: string;
+  /** Model that produced an assistant turn. */
+  model?: string;
+  /** Context chunks sent with this turn (assistant turns display them). */
+  sources?: ChatSource[];
+  attachments?: ChatAttachment[];
+  at: number;
+  /** Set when an assistant turn failed; content may hold a partial answer. */
+  error?: ChatError;
+};
+
+/**
+ * What the chat is allowed to ground on. `all` = every deck with notes,
+ * `none` = general chat, `decks` = a chosen set.
+ */
+export type ChatScope =
+  | { kind: "all" }
+  | { kind: "none" }
+  | { kind: "decks"; deckIds: string[] };
+
+export type ChatModelChoice = {
+  name: string;
+  badge: "text" | "vision" | "both";
+  available: boolean | null;
+};
+
+/** Last model choice, remembered across visits. */
+const MODEL_KEY = "freezmo.chat.model.v1";
+
+export function readStoredChatModel(): string | null {
+  try {
+    const value = window.localStorage.getItem(MODEL_KEY);
+    return value && value.trim() ? value : null;
+  } catch {
+    return null;
+  }
+}
+
+export function storeChatModel(model: string): void {
+  try {
+    window.localStorage.setItem(MODEL_KEY, model);
+  } catch {
+    /* private mode */
+  }
+}
+
+export function chatMessageId(prefix: string): string {
+  const rand =
+    typeof crypto !== "undefined" && "randomUUID" in crypto
+      ? crypto.randomUUID().slice(0, 8)
+      : Math.random().toString(36).slice(2, 10);
+  return `${prefix}_${Date.now().toString(36)}_${rand}`;
+}