Speeding up TwoGetUserMediaAndStop WebRTC test

As per the investigation in http://crbug.com/417756, it seems the Two*
tests are susceptible to timeouts on GPU-less, slow bots.

This patch will make the test no longer detect that video is up for
both streams before proceeding - we assume other tests check that.
Instead, we're happy with just getting a stream before proceeding.
This removes the costly video checks from this test so it can test
what it wants to test, which is to ensure we can stop streams in an
orderly fashion.

BUG=417756
R=perkj@chromium.org

Committed: https://crrev.com/8fb48317ec49417cf3257af945c6f3ada5c47311
Cr-Commit-Position: refs/heads/master@{#298884}

[phoglund_chromium, 2014-10-09 12:20:07]

https://codereview.chromium.org/640253002/diff/1/content/test/data/media/getu...
File content/test/data/media/getusermedia.html (right):

https://codereview.chromium.org/640253002/diff/1/content/test/data/media/getu...
content/test/data/media/getusermedia.html:169: var stream1 = null;
Unless of course you see a problem with doing the first gUMs in parallel
(although then I think our only option is to delete the test).

https://codereview.chromium.org/640253002/diff/1/content/test/data/media/getu...
content/test/data/media/getusermedia.html:249: failTest(result);
I've been annoyed by this rather bare error message, so I took the opportunity
to improve it.

[phoglund_chromium, 2014-10-09 13:07:46]

Actually, never mind, this actually makes the tests slower on Windows(!). I
really don't understand why.

[phoglund_chromium, 2014-10-09 13:27:03]

On 2014/10/09 13:07:46, phoglund wrote:
> Actually, never mind, this actually makes the tests slower on Windows(!). I
> really don't understand why.

Ok, I rewrote the stop test to start stopping streams without verifying that
stream 2 gets up. I still think it's quite safe, since we verify that stream 1
gets up later. I also noted that the stop test was actually fast enough, but
it's still nice to make it even faster (20 seconds on win is quite slow).

The aspect ratio detection is way too slow - those tests take upwards of 35
seconds to run on my win machine, so I'll try to fix those in my next patch.

[phoglund_chromium, 2014-10-09 13:59:27]

> The aspect ratio detection is way too slow - those tests take upwards of 35
> seconds to run on my win machine, so I'll try to fix those in my next patch.

Ok, I rewrote the patch. It no longer does gUM in parallel, since that made
things worse. This patch will now simply speed up the stop test.

I will make a separate patch to fix the other two* tests - the answer there is
too speed up the aspect ratio detector. It currently iterates 10 times, and that
takes ages on Windows. I'll try to make it iterate like once or twice instead.

[perkj_chrome, 2014-10-09 14:23:03]

lgtm

[phoglund_chromium, 2014-10-09 15:03:38]

The CQ bit was checked by phoglund@chromium.org

[commit-bot: I haz the power, 2014-10-09 15:04:36]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/640253002/120001

[commit-bot: I haz the power, 2014-10-09 16:07:06]

Committed patchset #7 (id:120001) as 8b0f306f10f06763a9875819fe84a0a21a5d18d8

[commit-bot: I haz the power, 2014-10-09 16:08:22]

Patchset 7 (id:??) landed as
https://crrev.com/8fb48317ec49417cf3257af945c6f3ada5c47311
Cr-Commit-Position: refs/heads/master@{#298884}