Replace uses of legacy SkBitmap::copyTo() API

chrome/browser/chromeos/accessibility/accessibility_highlight_manager_interactive_uitest.cc

Index: chrome/browser/chromeos/accessibility/accessibility_highlight_manager_interactive_uitest.cc
diff --git a/chrome/browser/chromeos/accessibility/accessibility_highlight_manager_interactive_uitest.cc b/chrome/browser/chromeos/accessibility/accessibility_highlight_manager_interactive_uitest.cc
index ae08305aba940c292907ab7a805d06327fb85ce2..70cc717c7f46bbe2d41578f2e22484b71e16c35e 100644
--- a/chrome/browser/chromeos/accessibility/accessibility_highlight_manager_interactive_uitest.cc
+++ b/chrome/browser/chromeos/accessibility/accessibility_highlight_manager_interactive_uitest.cc
@@ -81,12 +81,12 @@ class AccessibilityHighlightManagerTest : public InProcessBrowserTest {
 
   void CaptureBeforeImage(const gfx::Rect& bounds) {
     Capture(bounds);
-    image_.AsBitmap().deepCopyTo(&before_bmp_);
+    image_.asLegacyBitmap(&before_bmp_, SkImage::kRW_LegacyBitmapMode);

[f(malita), 2017/05/01 18:00:06]

asLegacyBitmap() doesn't copy the pixels for raster SkImages, so these are not
equivalent, no?

Depending on the test semantics, this may break.

[msarett1, 2017/05/01 18:04:55]

I think that using kRW_LegacyBitmapMode will force a copy.  Regardless, this
might be confusing.  What do you think?

[f(malita), 2017/05/01 18:36:01]

Ah yeah, indeed.  It does feel a bit fragile/incidental, but LGTM for a test.

Maybe add a comment to explain that we're forcing a deep copy.

[msarett1, 2017/05/01 19:53:50]

Agreed, yeah it is.  I've changed the impl to alloc() and readPixels()

   }
 
   void CaptureAfterImage(const gfx::Rect& bounds) {
     Capture(bounds);
-    image_.AsBitmap().deepCopyTo(&after_bmp_);
+    image_.asLegacyBitmap(&after_bmp_, SkImage::kRW_LegacyBitmapMode);

[f(malita), 2017/05/01 18:00:06]

Ditto.

   }
 
   void ComputeImageStats() {