Initial implementation of GeometryMapper

Initial implementation of GeometryMapper

Design document: https://docs.google.com/document/d/1WZKlOSUK4XI0Le0fgCsyUTVw0dTwutZXGWwzlHXewiU/edit

This patch implements the GeometryMapper class, with these methods:

LocalToVisibleRectInAncestorSpace: conceptually equivalent to LayoutObject::mapToVisualRectInAncestorSpace
LocalToAncestorRect: conceptually equivalent to LayoutObject::mapLocalToAncestor
AncestorToLocalRect: conceptually equivalent to LayoutObject::mapAncestorToLocal

Over time, this class plus the property tree builder will replace most of the use cases
of the above referenced methods.

Committed: https://crrev.com/9de57cf7d69d8b30d5ae66c902d6a9d501e68664
Cr-Commit-Position: refs/heads/master@{#401710}

[chrishtr, 2016-05-19 21:10:29]

Description was changed from

==========
none


none


none


none


none


none


none


none


none


none


none


none


none


none

BUG=
==========

to

==========
GeometryMapper
==========

[chrishtr, 2016-06-22 16:45:07]

Description was changed from

==========
GeometryMapper
==========

to

==========
Initial implementation of GeometryMapper

Design document:
https://docs.google.com/document/d/1WZKlOSUK4XI0Le0fgCsyUTVw0dTwutZXGWwzlHXew...

This patch implements the GeometryMapper class, with these methods:

LocalToVisibleRectInAncestorSpace: conceptually equivalent to
LayoutObject::mapToVisualRectInAncestorSpace

LocalToAncestorRect: conceptually equivalent to LayoutObject::mapLocalToAncestor

AncestorToLocalRect: conceptually equivalent to LayoutObject::mapAncestorToLocal

Over time, this class plus the property tree builder will replace most of the
use cases
of the above referenced methods.
==========

[chrishtr, 2016-06-22 21:57:00]

Description was changed from

==========
Initial implementation of GeometryMapper

Design document:
https://docs.google.com/document/d/1WZKlOSUK4XI0Le0fgCsyUTVw0dTwutZXGWwzlHXew...

This patch implements the GeometryMapper class, with these methods:

LocalToVisibleRectInAncestorSpace: conceptually equivalent to
LayoutObject::mapToVisualRectInAncestorSpace

LocalToAncestorRect: conceptually equivalent to LayoutObject::mapLocalToAncestor

AncestorToLocalRect: conceptually equivalent to LayoutObject::mapAncestorToLocal

Over time, this class plus the property tree builder will replace most of the
use cases
of the above referenced methods.
==========

to

==========
Initial implementation of GeometryMapper

Design document:
https://docs.google.com/document/d/1WZKlOSUK4XI0Le0fgCsyUTVw0dTwutZXGWwzlHXew...

This patch implements the GeometryMapper class, with these methods:

LocalToVisibleRectInAncestorSpace: conceptually equivalent to
LayoutObject::mapToVisualRectInAncestorSpace
LocalToAncestorRect: conceptually equivalent to LayoutObject::mapLocalToAncestor
AncestorToLocalRect: conceptually equivalent to LayoutObject::mapAncestorToLocal

Over time, this class plus the property tree builder will replace most of the
use cases
of the above referenced methods.
==========

[chrishtr, 2016-06-22 22:02:12]

chrishtr@chromium.org changed reviewers:
+ wangxianzhu@chromium.org, wkorman@chromium.org

[Xianzhu, 2016-06-22 23:20:31]

This is awesome!

Only comments about minor issues.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:53:
PrecomputedDataForAncestor* GeometryMapper::GetPrecomputedDataForAncestor(const
PropertyTreeState& ancestorState)
Can this return a reference?

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:57: return
&second->value;
It looks not clear what 'first' and 'second' are. How about writing them in one
statement?

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:60: const
FloatRect GeometryMapper::LocalToAncestorClipRect(
Return a reference?

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:105: const
PropertyTreeState& ancestorState) {
Return a reference?

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h (right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:19:
toAncestorTransforms;
Nit: The wrapping looks weird. I would put the above 2 lines into one line.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:51: // to or
a descendat of that in |ancestorState|.
Nit: s/descendat/descendant/.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:52: FloatRect
LocalToVisibleRectInAncestorSpace(const FloatRect&,
VisualRect?

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:56: // Maps
from a rect in |localTransformSpace}| to its transformed rect in
|ancestorSpace|. This is computed
Nit: extra '}'

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:74: private:
Nit: Insert a blank line before private:.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h:1: //
Copyright 2015 The Chromium Authors. All rights reserved.
Nit: 2016

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File
third_party/WebKit/Source/platform/graphics/paint/TransformPaintPropertyNode.h
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/TransformPaintPropertyNode.h:24:
static RefPtr<TransformPaintPropertyNode> create(const TransformationMatrix&
matrix, const FloatPoint3D& origin, PassRefPtr<TransformPaintPropertyNode>
parent = nullptr)
Why this change?

[wkorman, 2016-06-22 23:33:33]

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:55: auto
first = m_data.add(ancestorState.transform, PrecomputedDataForAncestor());
nit: believe this will always construct an empty PDFA which constructs two
HashMaps, which we only actually need to do if the value isn't already present.
Maybe this is canonical add-only-if-doesn't-already-exist usage for
wtf::HashMap, but could be worth check-first, construct/add-only-if-needed.

Doesn't this also create a copy, do we need to call std::move? There's some
precedent ex:

https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/layout/La...
https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/layout/Fl...

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:73: auto it
= precomputedData->toAncestorClipRects.find(clipNode);
There should only be zero or one, right? Can we use get() instead of an iterator
here and similar places in change?

https://cs.chromium.org/chromium/src/third_party/WebKit/Source/wtf/HashMap.h?...

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h (right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:40: // with
the same |ancestorState| parameter is guaranteed to be O(n + m),  where n is the
number of transform and clip/
rm '/' after clip

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:50: // It is
an error to call this method if any of the paint property tree nodes in
|localTransformState| not equal
are not equal

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:51: // to or
a descendat of that in |ancestorState|.
descendant (and 2x more below)

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:52: FloatRect
LocalToVisibleRectInAncestorSpace(const FloatRect&,
Visible -> Visual in method name?

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:60: // It is
an error to call this method if any of the paint property tree nodes in
|localTransformState| not equal
What would be returned in this case? Should we have success arg here as well?

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:66: // Maps
from a rect in |ancestorSpace}| to its transformed rect in
|localTransformSpace|. This is computed
rm } before |

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:71: FloatRect
AncestorToLocalRect(const FloatRect&,
Should briefly comment on success arg and return value in case where mapping
can't be done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h:1: //
Copyright 2015 The Chromium Authors. All rights reserved.
2016?

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h:17: struct
PropertyTreeState {
Worth adding a brief doc comment even if just noting it's a container for use
with GeometryMapper.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h:19:
TransformPaintPropertyNode* transformArg,
I think it's ok to name these without 'Arg' and they'll do the right thing in
the initializer list, see:

http://stackoverflow.com/a/6185043

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File
third_party/WebKit/Source/platform/graphics/paint/TransformPaintPropertyNode.h
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/TransformPaintPropertyNode.h:24:
static RefPtr<TransformPaintPropertyNode> create(const TransformationMatrix&
matrix, const FloatPoint3D& origin, PassRefPtr<TransformPaintPropertyNode>
parent = nullptr)
Clip and Effect headers are like this as well, should we change all of them?

[chrishtr, 2016-06-23 00:18:27]

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:53:
PrecomputedDataForAncestor* GeometryMapper::GetPrecomputedDataForAncestor(const
PropertyTreeState& ancestorState)
On 2016/06/22 at 23:20:30, Xianzhu wrote:
> Can this return a reference?

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:55: auto
first = m_data.add(ancestorState.transform, PrecomputedDataForAncestor());
On 2016/06/22 at 23:33:32, wkorman wrote:
> nit: believe this will always construct an empty PDFA which constructs two
HashMaps, which we only actually need to do if the value isn't already present.
Maybe this is canonical add-only-if-doesn't-already-exist usage for
wtf::HashMap, but could be worth check-first, construct/add-only-if-needed.
> 
> Doesn't this also create a copy, do we need to call std::move? There's some
precedent ex:
> 
>
https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/layout/La...
>
https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/layout/Fl...

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:57: return
&second->value;
On 2016/06/22 at 23:20:30, Xianzhu wrote:
> It looks not clear what 'first' and 'second' are. How about writing them in
one statement?

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:60: const
FloatRect GeometryMapper::LocalToAncestorClipRect(
On 2016/06/22 at 23:20:30, Xianzhu wrote:
> Return a reference?

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.cpp:73: auto it
= precomputedData->toAncestorClipRects.find(clipNode);
On 2016/06/22 at 23:33:32, wkorman wrote:
> There should only be zero or one, right? Can we use get() instead of an
iterator here and similar places in change?
> 
>
https://cs.chromium.org/chromium/src/third_party/WebKit/Source/wtf/HashMap.h?...

get returns a temporary according the compiler, and hence conflicts with
Xianzhu's request for a
reference return value. The latter seems more important..

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h (right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:19:
toAncestorTransforms;
On 2016/06/22 at 23:20:30, Xianzhu wrote:
> Nit: The wrapping looks weird. I would put the above 2 lines into one line.

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:40: // with
the same |ancestorState| parameter is guaranteed to be O(n + m),  where n is the
number of transform and clip/
On 2016/06/22 at 23:33:32, wkorman wrote:
> rm '/' after clip

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:50: // It is
an error to call this method if any of the paint property tree nodes in
|localTransformState| not equal
On 2016/06/22 at 23:33:32, wkorman wrote:
> are not equal

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:51: // to or
a descendat of that in |ancestorState|.
On 2016/06/22 at 23:33:32, wkorman wrote:
> descendant (and 2x more below)

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:52: FloatRect
LocalToVisibleRectInAncestorSpace(const FloatRect&,
On 2016/06/22 at 23:33:32, wkorman wrote:
> Visible -> Visual in method name?

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:56: // Maps
from a rect in |localTransformSpace}| to its transformed rect in
|ancestorSpace|. This is computed
On 2016/06/22 at 23:20:30, Xianzhu wrote:
> Nit: extra '}'

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:60: // It is
an error to call this method if any of the paint property tree nodes in
|localTransformState| not equal
On 2016/06/22 at 23:33:32, wkorman wrote:
> What would be returned in this case? Should we have success arg here as well?

Undefined. It should never happen and represents a bug in chrome. I have an
assert in the
implementation that checks for this situation.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:66: // Maps
from a rect in |ancestorSpace}| to its transformed rect in
|localTransformSpace|. This is computed
On 2016/06/22 at 23:33:32, wkorman wrote:
> rm } before |

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:71: FloatRect
AncestorToLocalRect(const FloatRect&,
On 2016/06/22 at 23:33:32, wkorman wrote:
> Should briefly comment on success arg and return value in case where mapping
can't be done.

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:74: private:
On 2016/06/22 at 23:20:30, Xianzhu wrote:
> Nit: Insert a blank line before private:.

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h:1: //
Copyright 2015 The Chromium Authors. All rights reserved.
On 2016/06/22 at 23:33:33, wkorman wrote:
> 2016?

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h:17: struct
PropertyTreeState {
On 2016/06/22 at 23:33:33, wkorman wrote:
> Worth adding a brief doc comment even if just noting it's a container for use
with GeometryMapper.

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/PropertyTreeState.h:19:
TransformPaintPropertyNode* transformArg,
On 2016/06/22 at 23:33:33, wkorman wrote:
> I think it's ok to name these without 'Arg' and they'll do the right thing in
the initializer list, see:
> 
> http://stackoverflow.com/a/6185043

Done.

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
File
third_party/WebKit/Source/platform/graphics/paint/TransformPaintPropertyNode.h
(right):

https://codereview.chromium.org/1973343003/diff/140001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/TransformPaintPropertyNode.h:24:
static RefPtr<TransformPaintPropertyNode> create(const TransformationMatrix&
matrix, const FloatPoint3D& origin, PassRefPtr<TransformPaintPropertyNode>
parent = nullptr)
On 2016/06/22 at 23:33:33, wkorman wrote:
> Clip and Effect headers are like this as well, should we change all of them?

Fixed.

[wkorman, 2016-06-23 00:41:10]

lgtm

https://codereview.chromium.org/1973343003/diff/160001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h (right):

https://codereview.chromium.org/1973343003/diff/160001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:43: // The
runtime of m calls among LocalToVisualeRectInAncestorSpace, LocalToAncestorRect
or AncestorToLocalRect
nit: typo 'e' after 'Visual'

[chrishtr, 2016-06-23 15:51:18]

https://codereview.chromium.org/1973343003/diff/160001/third_party/WebKit/Sou...
File third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h (right):

https://codereview.chromium.org/1973343003/diff/160001/third_party/WebKit/Sou...
third_party/WebKit/Source/platform/graphics/paint/GeometryMapper.h:43: // The
runtime of m calls among LocalToVisualeRectInAncestorSpace, LocalToAncestorRect
or AncestorToLocalRect
On 2016/06/23 at 00:41:10, wkorman wrote:
> nit: typo 'e' after 'Visual'

Done.

[chrishtr, 2016-06-23 15:51:21]

The CQ bit was checked by chrishtr@chromium.org

[chrishtr, 2016-06-23 15:51:21]

The patchset sent to the CQ was uploaded after l-g-t-m from wkorman@chromium.org
Link to the patchset: https://codereview.chromium.org/1973343003/#ps180001
(title: "none")

[commit-bot: I haz the power, 2016-06-23 15:51:36]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1973343003/180001

[commit-bot: I haz the power, 2016-06-23 16:43:56]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-06-23 16:43:57]

Try jobs failed on following builders:
  linux_chromium_asan_rel_ng on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[chrishtr, 2016-06-23 18:06:48]

The CQ bit was checked by chrishtr@chromium.org

[chrishtr, 2016-06-23 18:06:49]

The patchset sent to the CQ was uploaded after l-g-t-m from wkorman@chromium.org
Link to the patchset: https://codereview.chromium.org/1973343003/#ps220001
(title: "none")

[commit-bot: I haz the power, 2016-06-23 18:07:31]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1973343003/220001

[commit-bot: I haz the power, 2016-06-23 20:45:50]

Committed patchset #12 (id:220001)

[commit-bot: I haz the power, 2016-06-23 20:47:57]

Description was changed from

==========
Initial implementation of GeometryMapper

Design document:
https://docs.google.com/document/d/1WZKlOSUK4XI0Le0fgCsyUTVw0dTwutZXGWwzlHXew...

This patch implements the GeometryMapper class, with these methods:

LocalToVisibleRectInAncestorSpace: conceptually equivalent to
LayoutObject::mapToVisualRectInAncestorSpace
LocalToAncestorRect: conceptually equivalent to LayoutObject::mapLocalToAncestor
AncestorToLocalRect: conceptually equivalent to LayoutObject::mapAncestorToLocal

Over time, this class plus the property tree builder will replace most of the
use cases
of the above referenced methods.
==========

to

==========
Initial implementation of GeometryMapper

Design document:
https://docs.google.com/document/d/1WZKlOSUK4XI0Le0fgCsyUTVw0dTwutZXGWwzlHXew...

This patch implements the GeometryMapper class, with these methods:

LocalToVisibleRectInAncestorSpace: conceptually equivalent to
LayoutObject::mapToVisualRectInAncestorSpace
LocalToAncestorRect: conceptually equivalent to LayoutObject::mapLocalToAncestor
AncestorToLocalRect: conceptually equivalent to LayoutObject::mapAncestorToLocal

Over time, this class plus the property tree builder will replace most of the
use cases
of the above referenced methods.

Committed: https://crrev.com/9de57cf7d69d8b30d5ae66c902d6a9d501e68664
Cr-Commit-Position: refs/heads/master@{#401710}
==========

[commit-bot: I haz the power, 2016-06-23 20:47:59]

Patchset 12 (id:??) landed as
https://crrev.com/9de57cf7d69d8b30d5ae66c902d6a9d501e68664
Cr-Commit-Position: refs/heads/master@{#401710}