Phase 3 : chat streaming SSE et persistance des sessions

Backend :
- db.py : SQLite (sessions + messages), schéma et CRUD
- routes/sessions.py : créer/lister/ouvrir/renommer/supprimer une session
- routes/chat.py : POST /api/chat, relais token par token depuis Ollama (SSE),
  persistance des messages, titrage auto de la session au 1er message
- main.py : lifespan -> init_db, montage des nouvelles routes

Frontend :
- api/client.ts : APIs sessions + streamChat (parsing SSE event/data)
- store : sessions, messages, état de streaming, envoi optimiste
- ChatPanel : fil de conversation réel, bulles user/agent, curseur de frappe
- LeftPanel : historique réel (ouvrir/supprimer, horodatage relatif)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SVay7z3y7q2gEe54ByAE6N

frontend/src/api/client.ts

@@ -29,6 +29,99 @@ export async function getModels(): Promise<{
   return res.json();
 }
 
+export interface Session {
+  id: string;
+  title: string;
+  model?: string;
+  created_at: number;
+  updated_at: number;
+  message_count?: number;
+}
+
+export interface Message {
+  id: string;
+  session_id: string;
+  role: "user" | "assistant";
+  content: string;
+  model?: string;
+  created_at: number;
+}
+
+export async function listSessions(): Promise<Session[]> {
+  const res = await fetch("/api/sessions");
+  return (await res.json()).sessions;
+}
+
+export async function createSession(model?: string): Promise<Session> {
+  const res = await fetch("/api/sessions", {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({ title: "Nouvelle session", model }),
+  });
+  return res.json();
+}
+
+export async function getSession(
+  id: string
+): Promise<{ session: Session; messages: Message[] }> {
+  const res = await fetch(`/api/sessions/${id}`);
+  return res.json();
+}
+
+export async function deleteSession(id: string): Promise<void> {
+  await fetch(`/api/sessions/${id}`, { method: "DELETE" });
+}
+
+/** Envoie un message et streame la réponse de l'agent via SSE. */
+export async function streamChat(
+  body: { session_id: string; content: string; model?: string },
+  handlers: {
+    onToken: (t: string) => void;
+    onDone: (full: string) => void;
+    onError: (msg: string) => void;
+  }
+): Promise<void> {
+  const res = await fetch("/api/chat", {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify(body),
+  });
+  if (!res.body) {
+    handlers.onError("pas de flux de réponse");
+    return;
+  }
+
+  const reader = res.body.getReader();
+  const decoder = new TextDecoder();
+  let buffer = "";
+
+  while (true) {
+    const { done, value } = await reader.read();
+    if (done) break;
+    buffer += decoder.decode(value, { stream: true });
+    const events = buffer.split("\n\n");
+    buffer = events.pop() ?? "";
+
+    for (const block of events) {
+      let event = "message";
+      let data = "";
+      for (const line of block.split("\n")) {
+        if (line.startsWith("event: ")) event = line.slice(7).trim();
+        else if (line.startsWith("data: ")) data += line.slice(6);
+      }
+      if (!data) continue;
+      try {
+        const payload = JSON.parse(data);
+        if (event === "token") handlers.onToken(payload.content);
+        else if (event === "done") handlers.onDone(payload.content);
+        else if (event === "error") handlers.onError(payload.message);
+      } catch {
+        /* bloc partiel */
+      }
+    }
+  }
+}
+
 /** Télécharge un modèle en streamant la progression via SSE. */
 export async function pullModel(
   name: string,