test(cloudflare): Run the Think e2e suite serially with a 90s timeout

`getPlaywrightConfig` already defaults to `workers: 1` and a 30s timeout. The app
overrode both, so five tests each driving a live OpenRouter turn ran in parallel
under a cap meant for mocked models. That came from copying
`cloudflare-vercelai-v7`, which mocks its model; the two apps that call a real
provider, `cloudflare-mastra` and `node-flue`, keep the serial default and raise
the timeout to 90s instead.

Also corrects the comment above the first test, which still claimed the worker
imports no Sentry API. That stopped being true when the manual-span test was
added, and the worker now imports `startSpan` for exactly that.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Author: RulaKhaled

dev-packages/e2e-tests/test-applications/cloudflare-think/playwright.config.ts

@@ -13,10 +13,11 @@ const config = getPlaywrightConfig(
     startCommand: `pnpm preview`,
     port: APP_PORT,
   },
-  {
-    workers: '100%',
-    retries: 0,
-  },
+  // Each test drives a real OpenRouter tool-calling turn (up to two model calls) and then waits for
+  // the spans to flush, which does not fit the default 30s timeout when the provider is slow. The
+  // serial default from `getPlaywrightConfig` is kept: these turns share one worker and one event
+  // proxy, so running them in parallel only makes traces harder to tell apart.
+  { timeout: 90_000 },
 );
 
 export default config;

dev-packages/e2e-tests/test-applications/cloudflare-think/tests/think.test.ts

@@ -5,10 +5,11 @@ import { attr, isTurnOf, newAgentId, runAgentTurn, type StreamedSpan } from './u
 const APP = 'cloudflare-think';
 
 /**
- * The whole point of the app: the user's worker imports no Sentry API at all. `sentryCloudflareVitePlugin()`
+ * The whole point of the app: nothing wires Sentry into Think by hand. `sentryCloudflareVitePlugin()`
  * detects `class ThinkAgent extends Think` and wraps the export at build time, and the `ai` SDK publishes
- * the telemetry `vercelAIIntegration` consumes. Nothing here is Think-specific on the SDK side, so this
- * test is what would catch either half of that chain breaking.
+ * the telemetry `vercelAIIntegration` consumes. The worker's only Sentry import is the `startSpan` the
+ * manual-span test below needs. Nothing here is Think-specific on the SDK side, so this test is what
+ * would catch either half of that chain breaking.
  *
  * `collectStreamedSpans` rather than `waitForStreamedSpans`: the streamed `invoke_agent` parent stays
  * open until its children settle and flushes in a separate envelope from them, so a single envelope