webrtc: fix flaky TestStreamWriteDeadlineAsync

[sukunrt]

No description provided.

[marten-seemann]

I don't understand why this is necessary. We already have a 5x timeout on CI. Why is that not sufficient?

It seems like we have two ways here that are trying to achieve the same thing.

[sukunrt]

I'll rethink the whole test.

I'm not sure why we have the first extension of timeout in CI especially to a value as high as 500ms.

The test checks that when we set the deadline to timeout duration it does eventually give the deadline exceeded error. Then in the last check it checks that this error was provided after a sufficient duration. The last check defines sufficient currently as 1.5 * timeout. I changed it to 3 * timeout since it was flaking on windows as the test took more than 750 ms(500 * 1.5) and it generally took less than 900ms(as per my memory).
I am not convinced that the first extension of timeout makes sense. Why should we have to set deadline for 500ms when we are in CI?

[p-shahi]

What's the status of this PR - are we targeting it for the release or is it a nice to have?

[marten-seemann]

@Sukun says this test hasn't failed in a while. Closing.