Sporadic timeout when Puppeteer is launched

[tidoust]

Once in a while, the action fails when it launches Puppeteer due to a timeout, with the following error at the "Prepare" phase:

 [INFO] Finding shortName for Bikeshed document: video_frame_metadata_registry.src.html
    Timed out after 30000 ms while waiting for the WS endpoint URL to appear in stdout!

That happened 4 times in the past couple of months in the WebCodecs repository:
https://github.com/w3c/webcodecs/actions/runs/7693660342/job/20962850438
https://github.com/w3c/webcodecs/actions/runs/8164869456/job/22668028808
https://github.com/w3c/webcodecs/actions/runs/8150577505/job/22277066608
https://github.com/w3c/webcodecs/actions/runs/8267058031/job/22616549305

WebCodecs is a bit specific because a deploy job is 16 jobs. In these 4 examples, only one of these 16 jobs failed with a timeout. I thought this could be due to the jobs running in parallel but they run sequentially in practice, so that must be something else.

@dontcallmedom confirmed that he saw timeouts for WebRTC spec as well (he launches the job again when that happens, so no longer has the logs).

Not sure there's much that can be done about it. The timeout could be increased, of course, but then 30s seems more than enough to launch a browser.

Via w3c/webcodecs#782