Completely rewrite OpaqueBrowserFrameViewLayout unittests.

Completely rewrite OpaqueBrowserFrameViewLayout unittests.

The old tests simply hardcoded various size strings.  This was brittle against
changes, and did not work correctly in material design.

The new tests basically compute the important aspects of various bounds, without
simply calling into the real OBFVL positioning functions (which would make the
tests vacuous).  These are sufficiently parameterized to handle not only
material design but a variety of other state changes (maximized vs. restored,
hidden caption buttons, caption buttons on left, etc.).  This unfortunately
results in fairly complicated math, but frame positioning is complex, so there's
not much for it.

Many of these states were tested in sort of a one-off fashion, but the tests
have now been expanded to run a significantly more complete set of tests, in
particular testing just about everything in both restored and maximized mode.

As part of this rewrite, the tests now make use of some of the constants from
opaque_browser_frame_view_layout.cc, which have been exposed in the header for
this purpose.  Along the way, some of these were simplified a bit when we had
several constants that had the same value and served similar purposes.

BUG=575257
TEST=Unittests pass even with --top-chrome-md=material[-hybrid]

Committed: https://crrev.com/48ae591062e4c764d6f064e93956d1ae1a0e88fa
Cr-Commit-Position: refs/heads/master@{#371975}

[Peter Kasting, 2016-01-27 08:30:27]

pkasting@chromium.org changed reviewers:
+ sky@chromium.org

[Peter Kasting, 2016-01-27 08:30:27]

The test code is a bit impenetrable... sorry :(

[sky, 2016-01-27 17:58:20]

Rubber stamp LGTM - I would add that the way you have written this code errors
are going to take a bit of unraveling to figure out. In particular having loops
call off to functions with asserts means you don't know which iteration of the
loop is failing.

[Peter Kasting, 2016-01-28 02:02:25]

Adding tracing to indicate if windows are restored or maximized.  Shortened
EXPECT_ statements so failure messages would be more readable.  Simplified and
shortened delegate code.

Landing.

[Peter Kasting, 2016-01-28 02:03:11]

The CQ bit was checked by pkasting@chromium.org

[Peter Kasting, 2016-01-28 02:03:11]

The patchset sent to the CQ was uploaded after l-g-t-m from sky@chromium.org
Link to the patchset: https://codereview.chromium.org/1643453002/#ps20001
(title: "Rewrite")

[commit-bot: I haz the power, 2016-01-28 02:08:03]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1643453002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1643453002/20001

[commit-bot: I haz the power, 2016-01-28 02:55:06]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-01-28 02:55:54]

Description was changed from

==========
Completely rewrite OpaqueBrowserFrameViewLayout unittests.

The old tests simply hardcoded various size strings.  This was brittle against
changes, and did not work correctly in material design.

The new tests basically compute the important aspects of various bounds, without
simply calling into the real OBFVL positioning functions (which would make the
tests vacuous).  These are sufficiently parameterized to handle not only
material design but a variety of other state changes (maximized vs. restored,
hidden caption buttons, caption buttons on left, etc.).  This unfortunately
results in fairly complicated math, but frame positioning is complex, so there's
not much for it.

Many of these states were tested in sort of a one-off fashion, but the tests
have now been expanded to run a significantly more complete set of tests, in
particular testing just about everything in both restored and maximized mode.

As part of this rewrite, the tests now make use of some of the constants from
opaque_browser_frame_view_layout.cc, which have been exposed in the header for
this purpose.  Along the way, some of these were simplified a bit when we had
several constants that had the same value and served similar purposes.

BUG=575257
TEST=Unittests pass even with --top-chrome-md=material[-hybrid]
==========

to

==========
Completely rewrite OpaqueBrowserFrameViewLayout unittests.

The old tests simply hardcoded various size strings.  This was brittle against
changes, and did not work correctly in material design.

The new tests basically compute the important aspects of various bounds, without
simply calling into the real OBFVL positioning functions (which would make the
tests vacuous).  These are sufficiently parameterized to handle not only
material design but a variety of other state changes (maximized vs. restored,
hidden caption buttons, caption buttons on left, etc.).  This unfortunately
results in fairly complicated math, but frame positioning is complex, so there's
not much for it.

Many of these states were tested in sort of a one-off fashion, but the tests
have now been expanded to run a significantly more complete set of tests, in
particular testing just about everything in both restored and maximized mode.

As part of this rewrite, the tests now make use of some of the constants from
opaque_browser_frame_view_layout.cc, which have been exposed in the header for
this purpose.  Along the way, some of these were simplified a bit when we had
several constants that had the same value and served similar purposes.

BUG=575257
TEST=Unittests pass even with --top-chrome-md=material[-hybrid]

Committed: https://crrev.com/48ae591062e4c764d6f064e93956d1ae1a0e88fa
Cr-Commit-Position: refs/heads/master@{#371975}
==========

[commit-bot: I haz the power, 2016-01-28 02:55:55]

Patchset 2 (id:??) landed as
https://crrev.com/48ae591062e4c764d6f064e93956d1ae1a0e88fa
Cr-Commit-Position: refs/heads/master@{#371975}