chore(bidi): don't await browsingContext.close with runBeforeUnload

[hbenl]

Playwright expects page.close({ runBeforeUnload: true }) to resolve while the beforeunload dialog is still open but the BiDi command browsingContext.close only resolves after the dialog has been accepted or dismissed, so the implementation of page.close() should not await the response for browsingContext.close in this case.

Fixes:

• tests/library/beforeunload.spec.ts:
  • "should run beforeunload if asked for smoke"
  • "should access page after beforeunload"
• tests/library/headful.spec.ts:
  • "should close browsercontext with pending beforeunload dialog"

[github-actions]

Test results for "tests 1"

3 flaky

⚠️ [firefox-library] › library/browsercontext-reuse.spec.ts:114 › reuse launch › should reset serviceworker `@firefox-ubuntu-22.04-node18`
⚠️ [firefox-library] › library/inspector/cli-codegen-1.spec.ts:1079 › cli codegen › should not throw csp directive violation errors `@firefox-ubuntu-22.04-node18`
⚠️ [chromium-library] › library/inspector/cli-codegen-pick-locator.spec.ts:35 › should update locator highlight `@ubuntu-22.04-chromium-tip-of-tree`

32446 passed, 623 skipped

---

Merge workflow run.

[whimboo]

Please see #34281 when we as well tried to fix that. @juliandescottes do you remember where we ended-up with that?

[juliandescottes]

Not much to add compared to what can be read on the other PR. For the promptUnload bit, this PR is a quite similar to mine, but let's see.

BiDi spec issue is at w3c/webdriver-bidi#852 . We can put it to the agenda, for a next meeting to get feedback from the group. My opinion is that the library using the protocol is the best layer to actually decide whether to wait or not when promptUnload is true. The protocol being opinionated here wouldn't bring much added value.