Tentatively revert 268949 "Add browser_test for extension app API with missi..."

Tentatively revert 268949 "Add browser_test for extension app API with missi..."

It's suspected for breaking other browser tests. For more information,
see the failing win_chromium_rel try jobs in the code review!

> Add browser_test for extension app API with missing schema
> 
> This is a simplified version of api_test/stubs. It touches all the parts
> of each chrome.foo.bar.baz API path and ensures they are defined. This
> catches cases where a developer adding an API to _api_features.json but
> did not provide a schema for them.
> 
> BUG=369318
> TEST=added browser_tests ExtensionApiTest.StubsApp
> R=kalman@chromium.org
> 
> Review URL: https://codereview.chromium.org/264923009

TBR=jamescook@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=269046

[marja, 2014-05-08 10:14:43]

Committed patchset #1 manually as r269046 (tree was closed).

[marja, 2014-05-08 12:55:04]

This CL might not have been the problem, but I'm keeping this reverted for now.
There are many different types of failures at the moment, and it's hard to see
whether this was guilty to some of them or not. Some failures went away when I
reverted some commits...

(See crbug.com/371030.)

Sorry for probably unnecessary revert!

[James Cook, 2014-05-08 16:01:43]

On 2014/05/08 12:55:04, marja wrote:
> This CL might not have been the problem, but I'm keeping this reverted for
now.
> There are many different types of failures at the moment, and it's hard to see
> whether this was guilty to some of them or not. Some failures went away when I
> reverted some commits...
> 
> (See crbug.com/371030.)
> 
> Sorry for probably unnecessary revert!

I know about crbug.com/371030 -- I filed it. Those tests were failing both with
and without my patch, so I don't think they were related to my code.  My code
only adds a new test and does not touch any of the Chrome implementation code.

[marja, 2014-05-08 16:03:32]

Yeah, the theory where this commit could've caused something was paranoid. (It
was related to the failure only occurring when sharding, and sharding runs the
tests in single process mode, so tests could theoretically interfere with other
tests...) But now it looks even worse, basically, there's no known good revision
any more :/

[James Cook, 2014-05-08 16:05:03]

On 2014/05/08 16:03:32, marja wrote:
> Yeah, the theory where this commit could've caused something was paranoid. (It
> was related to the failure only occurring when sharding, and sharding runs the
> tests in single process mode, so tests could theoretically interfere with
other
> tests...) But now it looks even worse, basically, there's no known good
revision
> any more :/

No worries, this patch isn't important and can wait.  Sometimes you gotta just
keep reverting. :-)

[marja, 2014-05-09 07:41:16]

... and it turned out to be this patch after all!

The world is weird and the most paranoid theories turn out to be correct...