chore(release): v0.4.5 — surface "Restart server to apply" hint on QUANT/CONTEXT/SERVER saves

User-reported issue: changing TurboQuant mode (or any other env-derived
knob) in the QUANT card silently no-ops on the live server because
`update_quant_config` / `update_context_config` / `update_server_config`
only persist to disk — `lumen-server`'s env vars are baked at spawn time
(see `server.rs::spawn_server_command`) and don't re-read the config.

Previously only the ENV OVERRIDES card showed
  "Saved. Restart the server to apply (Stop → Start)."
The QUANT, CONTEXT, and SERVER cards saved silently — `saveQuant()` and
`saveContext()` didn't even set `statusMessage`, and `saveServer()`
showed a generic "Server config saved" without the restart hint.
Users had no way to know their TurboQuant / KV-quant / ctx caps weren't
actually live.

This patch:
  * Two new i18n keys (EN + KO):
    - `config.savedRestartHint` — "Saved. Restart the server to apply
      (Stop → Start)."
    - `config.saved` — bare "Saved." for when the server is stopped
      (next start picks up the change naturally — no restart needed).
  * `savedToast()` helper in App.svelte picks the right variant by
    checking `status.state in {"running", "starting"}`.
  * `saveServer()`, `saveQuant()`, `saveContext()` all invoke
    `savedToast()` after the on-disk save + clear after 3s (up from
    1.5s — the longer text needs more dwell time to read).

No behavior change to the server itself — pure UX surfacing. Users on
v0.4.4 who hit the silent-save trap should see the hint immediately
after OTA update + opening any of the three cards.

Author: rabbitson87

Cargo.lock

@@ -1,6 +1,6 @@
 [package]
 name = "lumen-app"
-version = "0.4.4"
+version = "0.4.5"
 edition.workspace = true
 rust-version.workspace = true
 license.workspace = true

crates/lumen-app/Cargo.toml

@@ -1,7 +1,7 @@
 {
   "name": "lumen-app-frontend",
   "private": true,
-  "version": "0.4.4",
+  "version": "0.4.5",
   "type": "module",
   "scripts": {
     "predev": "cargo build --manifest-path ../../../Cargo.toml -p lumen-server --release",

crates/lumen-app/frontend/package.json

@@ -435,16 +435,32 @@
     config = await api.updateServerConfig(config.server);
   }
 
+  // Shared toast text — restart hint only when the server is currently
+  // running (otherwise just "Saved", since next start will pick the new
+  // env vars naturally). Without this, users who change TurboQuant /
+  // KV-quant / ctx caps while the server is running see the UI update
+  // but the running process keeps the OLD env vars until manual
+  // Stop → Start. The QUANT / CONTEXT / SERVER cards previously had no
+  // hint at all (only env_overrides did).
+  function savedToast(): string {
+    if (status.state === "running" || status.state === "starting") {
+      return t("config.savedRestartHint");
+    }
+    return t("config.saved");
+  }
+
   async function saveServer() {
     if (!config) return;
     config = await api.updateServerConfig(config.server);
-    statusMessage = "Server config saved";
-    setTimeout(() => (statusMessage = null), 1500);
+    statusMessage = savedToast();
+    setTimeout(() => (statusMessage = null), 3000);
   }
 
   async function saveQuant() {
     if (!config) return;
     config = await api.updateQuantConfig(config.quant);
+    statusMessage = savedToast();
+    setTimeout(() => (statusMessage = null), 3000);
   }
 
   async function saveContext() {
@@ -453,6 +469,8 @@
     // ctx affects KV-cache headroom in the tuned memory recommendation
     // (~1 GB per 8K tokens). Re-sync so saved caps follow.
     await syncTunedMemoryCaps();
+    statusMessage = savedToast();
+    setTimeout(() => (statusMessage = null), 3000);
   }
 
   async function resetMemoryCaps() {

crates/lumen-app/frontend/src/App.svelte

@@ -276,6 +276,13 @@ export const en: Record<string, string> = {
   "env.applyHint": "Changes apply on next server start.",
   "env.empty2": "No overrides set.",
 
+  // Shared toast for QUANT / CONTEXT / SERVER card saves. The variant shown
+  // depends on whether the inference server is currently running — running
+  // server needs a restart for env-derived knobs (TurboQuant, KV quant, ctx
+  // caps, etc.) to take effect.
+  "config.savedRestartHint": "Saved. Restart the server to apply (Stop → Start).",
+  "config.saved": "Saved.",
+
   // ── DoctorPanel ─────────────────────────────────────────────────
   "doctor.title": "Doctor",
   "doctor.run": "Run checks",

crates/lumen-app/frontend/src/messages/en.ts

@@ -267,6 +267,12 @@ export const ko: Record<string, string> = {
   "env.applyHint": "변경 사항은 다음 서버 시작 시 적용됩니다.",
   "env.empty2": "설정된 오버라이드가 없습니다.",
 
+  // QUANT / CONTEXT / SERVER 카드 저장 시 공통 토스트. 서버 실행 중이면
+  // 재시작 안내, 정지 상태면 단순 "저장됨" 만 표시. env-derived 노브
+  // (TurboQuant, KV quant, ctx caps 등) 는 재시작 시점에만 반영됨.
+  "config.savedRestartHint": "저장됨. 적용하려면 서버 재시작 (중지 → 시작).",
+  "config.saved": "저장됨.",
+
   // ── 진단 패널 ───────────────────────────────────────────────────
   "doctor.title": "진단",
   "doctor.run": "점검 실행",

crates/lumen-app/frontend/src/messages/ko.ts

@@ -1,7 +1,7 @@
 {
   "$schema": "https://schema.tauri.app/config/2",
   "productName": "Lumen",
-  "version": "0.4.4",
+  "version": "0.4.5",
   "identifier": "ai.lumen.app",
   "build": {
     "frontendDist": "frontend/dist",