Fix tests to be compatible with Android scale initialization.

Fix tests to be compatible with Android scale initialization.

On Android, we never want to allow zooming out further than the
content size.  This behavior is currently guarded by the shrinksViewportContentToFit
WebSetting.  Fix the layout tests to be compatible with this setting in
the future by expanding their body width.  I also removed the viewportEnabled
check to activate the behavior, which was redundant.

Unit tests also need some changes to be compatible with the behavior 
to always zoom out to minimum scale on page load unless 
setPageScaleFactorLimits is called (which it always is in non-Android
content_shell).  I fixed most of them to call setPageScaleFactorLimits
and deleted WebFrameTest.setPageScaleFactorBeforeFrameHasView (which
made no sense).

These changes will help prepare for better convergence of scaling behavior
between platforms in the future.

NOTRY=true
BUG=350599
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=173979

[aelias_OOO_until_Jul13, 2014-03-08 01:39:32]

Hi bokan, PTAL.  This is just a bit of cleanup I've been wanting to do for
awhile.

[bokan, 2014-03-10 14:21:17]

LGTM, thanks!

https://codereview.chromium.org/190723007/diff/60001/Source/web/tests/WebFram...
File Source/web/tests/WebFrameTest.cpp (left):

https://codereview.chromium.org/190723007/diff/60001/Source/web/tests/WebFram...
Source/web/tests/WebFrameTest.cpp:1567: TEST_F(WebFrameTest,
setPageScaleFactorBeforeFrameHasView)
Is the reason this doesn't make sense because page scale factor is only set by
the user or from the page itself so it must be after the frame is loaded?

[aelias_OOO_until_Jul13, 2014-03-10 21:43:42]

https://codereview.chromium.org/190723007/diff/60001/Source/web/tests/WebFram...
File Source/web/tests/WebFrameTest.cpp (left):

https://codereview.chromium.org/190723007/diff/60001/Source/web/tests/WebFram...
Source/web/tests/WebFrameTest.cpp:1567: TEST_F(WebFrameTest,
setPageScaleFactorBeforeFrameHasView)
On 2014/03/10 14:21:18, bokan wrote:
> Is the reason this doesn't make sense because page scale factor is only set by
> the user or from the page itself so it must be after the frame is loaded?

Right.  Early-ish in the load (after contents width is obtained) the page scale
is clobbered by the minimum scale, and then possibly by the history item, so
there's no scenario where we would preserve the previous page scale.  This test
only passed before because it didn't enable the viewport tag or set an initial
scale override.

[aelias_OOO_until_Jul13, 2014-05-13 02:47:45]

Resurrecting this patch as a test-only change.  I never landed it because a
mac_blink_rel test failure pointed to a bug with non-overlay scrollbars (where
scrollbar size and view size are updated at different times), which will likely
cause scale to get initialized to >1 on desktop in some scenarios.  I never did
have time to dig into that bug and still don't; but the test cleanups in this
patch are desirable anyway, so I'm planning to land only those now.

[aelias_OOO_until_Jul13, 2014-05-13 18:48:43]

The CQ bit was checked by aelias@chromium.org

[commit-bot: I haz the power, 2014-05-13 18:49:09]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/aelias@chromium.org/190723007/200001

[commit-bot: I haz the power, 2014-05-13 18:51:12]

Change committed as 173979

[kenneth.christiansen, 2014-06-24 14:03:55]

On 2014/05/13 18:51:12, I haz the power (commit-bot) wrote:
> Change committed as 173979

Not all of these tests are actually working if you turn on that setting. Like
you cannot set the size of a frameset.

The broken tests are

+fast/frames/frame-set-rotation-hit.html
+fast/frames/frame-set-scaling-hit.html

I don't know how to intent to fix them as, at least the former, depends on being
able to scale the frameset down. I don't really understand the test though

The comment in the test is: // Scale the page and resize the frames to verify
hit testing on the frameset

In a window of 800x600 it creates a frameset of 25,75% (200px, 600px) which it
then rotates 90 degrees in the middle and translates down 100px to start from
the top of the screen.

Meaning we now have 100px white space in both sides and from the top to the
button, 200px height frame1 followed by 600px height frame2. This is the result
that I get with the setting enabled.

Now after the scale adjustment (0.5 in 0,0), and the resulting image is scaled
but the frameset keeps its original size, ie you see frame1 still as 200px
height and then you only see 200px of frame2. 

I don't get why that is even correct (like iOS supported (still does?) frameset
and (with frame flattening) pinch zooming would just scale the frameset along.