perf(relay): full-chain optimization — key pool, model sync, SSE stream

Phase 1 (Key Pool correctness):
- RPM: fixed-minute window → sliding 60s aggregation (prevents 2x burst)
- Remove fallback-to-provider-key bypass when all keys rate-limited
- SSE semaphore: 16→64 permits, cleanup delay 60s→5s
- Default 429 cooldown: 5min→60s (better for Coding Plan quotas)
- Expire old key_usage_window rows on record

Phase 2 (Frontend model sync):
- currentModel empty-string fallback to glm-4-flash-250414 in relay client
- Merge duplicate listModels() calls in connectionStore SaaS path
- Show ModelSelector in Tauri mode when models available
- Clear currentModel on SaaS logout

Phase 3 (Relay performance):
- Key Pool: DashMap in-memory cache (TTL 5s) for select_best_key
- Cache invalidation on 429 marking

Phase 4 (SSE stream):
- AbortController integration for user-cancelled streams
- SSE parsing: split by event boundaries (\n\n) instead of per-line
- streamStore cancelStream adapts to 0-arg and 1-arg cancel fns

desktop/src/store/chat/streamStore.ts

@@ -581,11 +581,20 @@ export const useStreamStore = create<StreamState>()(
     if (!isStreaming) return;
 
     // 1. Tell backend to abort — use sessionKey (which is the sessionId in Tauri)
+    //    Also abort the frontend SSE fetch via cancelStream()
     try {
-      const client = getClient();
+      const client = getClient() as unknown as Record<string, unknown>;
       if ('cancelStream' in client) {
-        const sessionId = useConversationStore.getState().sessionKey || activeRunId || '';
-        (client as { cancelStream: (id: string) => void }).cancelStream(sessionId);
+        const fn = client.cancelStream;
+        if (typeof fn === 'function') {
+          // Call with or without sessionId depending on arity
+          if (fn.length > 0) {
+            const sessionId = useConversationStore.getState().sessionKey || activeRunId || '';
+            (fn as (id: string) => void)(sessionId);
+          } else {
+            (fn as () => void)();
+          }
+        }
       }
     } catch {
       // Backend cancel is best-effort; proceed with local cleanup