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/components/TopBar.tsx

@@ -1,8 +1,9 @@
 import { useEffect, useState } from "react";
 import { useStore } from "../store/useStore";
 import { getVersion } from "../api/client";
-import { ChevronDown, LokiMark } from "./Icon";
+import { LokiMark } from "./Icon";
 import { ModelSelector } from "./ModelSelector";
+import { SessionMenu } from "./SessionMenu";
 
 /** Barre supérieure sombre : logo, fil d'Ariane, statut Ollama, sélecteur. */
 export function TopBar() {
@@ -33,10 +34,7 @@ export function TopBar() {
 
       <div className="h-6 w-[3px] bg-chrome-3" />
 
-      <div className="flex items-center gap-2 text-[13px] text-on-dark">
-        <span>Nouvelle session</span>
-        <ChevronDown className="text-on-dark-3" />
-      </div>
+      <SessionMenu />
 
       <div className="flex-1" />