Snippet in article about setSinkId() may cause a memory leak

[chrisguttandin]

Describe the bug
The article about setSinkId() (https://developer.chrome.com/en/blog/audiocontext-setsinkid/) contains a snippet which may cause a memory leak.

The article recommends to trigger the permission to view all audio output devices by calling getUserMedia() like this:

await navigator.mediaDevices.getUserMedia({ audio: true });

As far as I can tell this will keep the returned stream alive indefinitely in Chrome (tested in v109 and v111). Firefox for example will garbage collect the stream.

To Reproduce
Steps to reproduce the behavior:

1. Run the line of code from above.
2. Observe that the microphone indicator of the operating system (tested on macOS) will never go away again.

Expected behavior
I think the stream should be stopped immediately when getUserMedia() is only used to trigger the permission to view more devices.

const mediaStream = await navigator.mediaDevices.getUserMedia({ audio: true });

mediaStream.getTracks().forEach((track) => track.stop());

I'm happy to provide a PR to update the article.

Additional context
By coincidence there is currently a spec discussion if garbage collecting an unused stream like that should be allowed or not. w3c/mediacapture-main#910

[beaufortfrancois]

You're absolutely right @chrisguttandin
Thank you for raising this issue. I've started #4982 to address it.