fix(electron): stalling on delayed process close

[mxschmitt]

Investigation notes:

When we call ElectronApplication.close(), we return before the actual process has been exited. This can/will lead under certain timing conditions to that gracefullyCloseSet still has an entry.
When the worker then tries to exit, we call these entries, which will do options.attemptToGracefullyClose(). A recursive call to ElectronApplication.close() which will never resolve, due to the fact that we call BrowserContext.close again (good) but the actual close event will only be fired once. See here. This is why it was stalling for 30s.

Changes:

Based on that, I see 2 changes which we should do:

1. Adjust ElectronApplication.close() so it can be called multiple times (like for a normal browser)
2. Wait until the process has exited before returning from ElectronApplication.close() (like for a normal browser - Browser.close() -> gracefullyClose).

Fixes #27523

[github-actions]

Test results for "tests 1"

7 flaky

⚠️ [chromium] › library/tracing.spec.ts:243:5 › should not include trace resources from the previous chunks
⚠️ [chromium] › library/tracing.spec.ts:243:5 › should not include trace resources from the previous chunks
⚠️ [chromium] › library/tracing.spec.ts:243:5 › should not include trace resources from the previous chunks
⚠️ [playwright-test] › ui-mode-test-ct.spec.ts:215:5 › should watch component via util
⚠️ [chromium] › library/tracing.spec.ts:243:5 › should not include trace resources from the previous chunks
⚠️ [webkit] › library/tracing.spec.ts:243:5 › should not include trace resources from the previous chunks
⚠️ [playwright-test] › ui-mode-test-ct.spec.ts:59:5 › should run component tests after editing test

26581 passed, 611 skipped
✔️✔️✔️

Merge workflow run.