fix(openai): preserve GPT-Live startup after recoverable provider errors (#155534)

Related: #155190 (provider-startup portion only; telephone-call teardown remains separate).

## What Problem This Solves

Fixes GPT-Live sessions failing to start when a recoverable provider error arrives before `session.started`.

## User Impact

Recoverable warnings no longer prevent a subsequent readiness event from starting the voice session. Fatal authentication errors still reject startup immediately. No configuration or migration changes are needed.

## Why This Change Was Made

The bridge now checks the existing fatal-auth classification before rejecting startup, logs a redacted warning for recoverable startup errors, and continues waiting within the existing readiness timeout. Established-session error behavior and redaction remain unchanged. The classifier is unchanged: status 401 (number or string), or codes `authentication_error`, `invalid_api_key`, `invalid_token`, and `token_expired` are fatal; other error events, including `missing_scope` without status 401, are nonfatal.

Provider-error dispatch stays in a small private sibling module because the bridge is already at its enforced line cap. The regression checks readiness and terminal behavior rather than exact warning wording.

## Evidence

- Real bridge `connect()` replay using a real `ws` client and a TCP loopback WebSocket server, production event parsing/lifecycle, and the existing media-runtime test adapter. The server sent `error` followed by `session.started` after receiving `session.start`.
- Baseline `4d50b52bce`: `missing_scope` rejected startup; connected=false, ready callbacks=0. Candidate: startup resolved; connected=true, ready callbacks=1, warning=1, error callbacks=0, close callbacks=0.
- `authentication_error` rejected startup on both baseline and candidate, with connected=false and ready callbacks=0 despite the subsequent readiness frame.
- `node scripts/run-vitest.mjs extensions/openai/realtime-quicksilver-bridge.test.ts --maxWorkers=1`: 37 passed; measured command wall time 38.71 seconds (Vitest 34.73 seconds). Earlier exact contributor-head CI changed-extension test-shard step took 133 seconds.
- Replay ran under Bun 1.4.2; this is synthetic provider-stream transport proof, not live OpenAI availability, full Gateway, telephone-call, or media-worker packaging proof. No real provider credentials were used.

Co-authored-by: Ayaan Zaidi <hi@obviy.us>

extensions/openai/realtime-quicksilver-bridge.test.ts

@@ -478,6 +478,29 @@ describe("OpenAIQuicksilverVoiceBridge", () => {
     expect(harness.onReady).toHaveBeenCalledOnce();
   });
 
+  it("keeps recoverable provider errors non-terminal before session readiness", async () => {
+    const harness = createHarness({ autoStart: false });
+    const connecting = harness.bridge.connect();
+    await vi.waitFor(() => expect(harness.socket.readyState).toBe(1));
+
+    harness.socket.serverEvent({
+      type: "error",
+      error: { code: "missing_scope", message: "temporary provider rejection" },
+    });
+
+    expect(harness.bridge.isConnected()).toBe(false);
+    expect(harness.onError).not.toHaveBeenCalled();
+    expect(harness.onClose).not.toHaveBeenCalled();
+
+    harness.socket.serverEvent({
+      type: "session.started",
+      session: { id: "live-1", expires_at: Math.floor(Date.now() / 1000) + 60 },
+    });
+    await connecting;
+    expect(harness.bridge.isConnected()).toBe(true);
+    expect(harness.onReady).toHaveBeenCalledOnce();
+  });
+
   it.each([
     [1000, "completed"],
     [1006, "error"],
@@ -595,7 +618,7 @@ describe("OpenAIQuicksilverVoiceBridge", () => {
 
     harness.socket.serverEvent({
       type: "error",
-      error: { message: "invalid live session" },
+      error: { code: "authentication_error", message: "invalid live session" },
     });
 
     await expect(connecting).rejects.toThrow("OpenAI GPT-Live transport failed");

extensions/openai/realtime-quicksilver-bridge.ts

@@ -21,6 +21,7 @@ import {
   closeOpenAILiveSocket,
   openAIQuicksilverToolResultText,
 } from "./realtime-quicksilver-protocol.js";
+import { handleOpenAIQuicksilverProviderError } from "./realtime-quicksilver-provider-error.js";
 import { projectOpenAIQuicksilverErrorMessage } from "./realtime-quicksilver-redaction.js";
 import {
   connectOpenAIQuicksilverSideband,
@@ -562,18 +563,17 @@ export class OpenAIQuicksilverVoiceBridge implements RealtimeVoiceBridge {
     }
     const message = projectOpenAIQuicksilverErrorMessage("provider");
     const error = new Error(message);
-    if (!this.lifecycle.isReady()) {
-      failStartup(error, "session start failed");
-      return;
-    }
     const reportEvent = () =>
       this.config.onEvent?.({ direction: "server", type: "error", detail: message });
-    if (event.fatalAuth) {
-      this.fail(connection, error, "authentication failed", reportEvent);
-    } else {
-      reportEvent();
-      this.config.onError?.(error);
-    }
+    handleOpenAIQuicksilverProviderError({
+      fatalAuth: event.fatalAuth,
+      ready: this.lifecycle.isReady(),
+      failStartup: () => failStartup(error, "session start failed"),
+      failAuthentication: () => this.fail(connection, error, "authentication failed", reportEvent),
+      reportEvent,
+      reportError: () => this.config.onError?.(error),
+      logger: this.config.logger,
+    });
   }
 
   private startDelegation(

extensions/openai/realtime-quicksilver-provider-error.ts

@@ -0,0 +1,38 @@
+import type { PluginLogger } from "openclaw/plugin-sdk/plugin-entry";
+
+type OpenAIQuicksilverProviderErrorHandling = {
+  fatalAuth: boolean;
+  ready: boolean;
+  failStartup: () => void;
+  failAuthentication: () => void;
+  reportEvent: () => void;
+  reportError: () => void;
+  logger?: Pick<PluginLogger, "warn">;
+};
+
+export function handleOpenAIQuicksilverProviderError({
+  fatalAuth,
+  ready,
+  failStartup,
+  failAuthentication,
+  reportEvent,
+  reportError,
+  logger,
+}: OpenAIQuicksilverProviderErrorHandling): void {
+  if (fatalAuth) {
+    if (!ready) {
+      failStartup();
+    } else {
+      failAuthentication();
+    }
+    return;
+  }
+  reportEvent();
+  if (!ready) {
+    (logger?.warn ?? console.warn)(
+      "OpenAI GPT-Live provider error before session startup; continuing readiness",
+    );
+    return;
+  }
+  reportError();
+}