Add desired width & height to drawContext (as opposed to using the width & height of the RT)

Add desired width & height to drawContext (as opposed to using the width & height of the RT)

This is split out of: https://codereview.chromium.org/2168483003/ (Deprecate SkDevice::accessBitmap method). The GPUDevice can no longer rely on the fLegacyBitmap to store the device's width & height

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2163323002

[robertphillips, 2016-07-20 18:07:20]

Description was changed from

==========
Add desired width & height to drawContext (as opposed to using the width &
height of the RT)
==========

to

==========
Add desired width & height to drawContext (as opposed to using the width &
height of the RT)
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2163323002
==========

[robertphillips, 2016-07-20 18:08:35]

Description was changed from

==========
Add desired width & height to drawContext (as opposed to using the width &
height of the RT)
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2163323002
==========

to

==========
Add desired width & height to drawContext (as opposed to using the width &
height of the RT)

This is split out of: https://codereview.chromium.org/2168483003/ (Deprecate
SkDevice::accessBitmap method). The GPUDevice can no longer rely on the
fLegacyBitmap to store the device's width & height

GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2163323002
==========

[robertphillips, 2016-07-20 18:08:35]

robertphillips@google.com changed reviewers:
+ bsalomon@google.com

[robertphillips, 2016-07-20 20:00:40]

The CQ bit was checked by robertphillips@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-07-20 20:00:51]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-20 20:25:07]

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

[commit-bot: I haz the power, 2016-07-20 20:25:07]

Dry run: This issue passed the CQ dry run.

[bsalomon, 2016-07-21 13:43:43]

Hm, why does the draw context need to know which portion to client is rendering
into?

[robertphillips, 2016-07-21 13:49:07]

On 2016/07/21 13:43:43, bsalomon wrote:
> Hm, why does the draw context need to know which portion to client is
rendering
> into?

In https://codereview.chromium.org/2168483003/ (Deprecate SkDevice::accessBitmap
method) I just add a new fSize field to SkGpuDevice. You can see if you would
just like to go with that for now.

We will have to deal with a ~related issue when SkGpuDevice's drawContext is
actually backed by a RenderTargetProxy though. The RTP will be storing the
desired size of the RenderTarget, not the allocated size. All the locations
where we need to peer in and get the allocated size will need to be deferred
until the RTP is actually backed by a RenderTarget.

[robertphillips, 2016-07-21 14:49:16]

Moth-balling for now. Will revisit when drawContext's are RenderTargetProxy
backed. Hopefully it will make more sense then.