perf: fix model reload thrash and cut time-to-first-token

Ollama reloaded the chat model mid-message because plan/summary/router
calls sent divergent runner options (no num_ctx/num_batch) and omitted
keep_alive — the main cause of perceived slowness.

- share one keep-alive httpx.AsyncClient for all Ollama calls
- unify runner options (runner_options) + keep_alive on every model
  call, embeddings included
- drop the blocking LLM routing fallback (pure lexical heuristic)
- run RAG recall + plan + code-model pick in parallel inside the SSE
  stream, after the start event
- RAG cosine scoring off the event loop; cache /api/tags 30s and
  nvidia-smi 5s; frontend polls 2s->5s, warm poll backoff, dedup
  config fetch

feat: working session menu in TopBar (switch/create/rename/delete)
feat: workspace file deletion (DELETE /api/files + UI trash buttons)
docs: recommended Ollama env vars (KEEP_ALIVE, MAX_LOADED_MODELS...)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

frontend/src/store/useStore.ts

@@ -1,6 +1,7 @@
 import { create } from "zustand";
 import {
   createSession,
+  deleteFile,
   deleteSession,
   getConfig,
   getModels,
@@ -12,6 +13,7 @@ import {
   fileContent,
   listFiles,
   listSessions,
+  renameSession as apiRenameSession,
   runShell,
   saveConfig,
   streamChat,
@@ -73,6 +75,8 @@ interface LokiState {
   newSession: () => Promise<void>;
   openSession: (id: string) => Promise<void>;
   removeSession: (id: string) => Promise<void>;
+  renameSession: (id: string, title: string) => Promise<void>;
+  removeFile: (path: string) => Promise<void>;
   sendMessage: (content: string) => Promise<void>;
   stopStreaming: () => void;
 
@@ -278,6 +282,23 @@ export const useStore = create<LokiState>((set, get) => ({
     await get().refreshSessions();
   },
 
+  renameSession: async (id, title) => {
+    const clean = title.trim();
+    if (!clean) return;
+    await apiRenameSession(id, clean);
+    await get().refreshSessions();
+  },
+
+  removeFile: async (path) => {
+    await deleteFile(path);
+    // Ferme l'aperçu si le fichier supprimé (ou son dossier) y est affiché.
+    const preview = get().previewPath;
+    if (preview && (preview === path || preview.startsWith(path + "/"))) {
+      set({ previewPath: null, previewContent: "" });
+    }
+    await get().refreshFiles();
+  },
+
   stopStreaming: () => {
     activeStreamController?.abort();
     activeStreamController = null;
@@ -403,10 +424,15 @@ export const useStore = create<LokiState>((set, get) => ({
             streamNotice: null,
             streamTools: [],
           });
-          // Recharge depuis la base + l'arborescence (fichiers créés par l'agent).
-          if (get().currentSessionId === sid) await get().openSession(sid!);
-          await get().refreshSessions();
-          await get().refreshFiles();
+          // Recharge depuis la base + l'arborescence (fichiers créés par
+          // l'agent) — trois requêtes indépendantes, en parallèle.
+          await Promise.all([
+            get().currentSessionId === sid
+              ? get().openSession(sid!)
+              : Promise.resolve(),
+            get().refreshSessions(),
+            get().refreshFiles(),
+          ]);
           if (writtenHtml) await get().openPreview(writtenHtml.args.path as string);
         },
         onError: (msg) => {