Don't use a SubtreeLayoutScope object for SVG resource layout

third_party/WebKit/LayoutTests/svg/custom/patterns-nested-no-geometry.svg

Index: third_party/WebKit/LayoutTests/svg/custom/patterns-nested-no-geometry.svg
diff --git a/third_party/WebKit/LayoutTests/svg/custom/patterns-nested-no-geometry.svg b/third_party/WebKit/LayoutTests/svg/custom/patterns-nested-no-geometry.svg
new file mode 100644
index 0000000000000000000000000000000000000000..861a8ec98b12d0e4ac12c838a4b30fd5d04874e7
--- /dev/null
+++ b/third_party/WebKit/LayoutTests/svg/custom/patterns-nested-no-geometry.svg
@@ -0,0 +1,8 @@
+<svg width="100%" height="100%" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
+  <rect fill="url(#p1)" stroke="green" x="10" y="10" width="50" height="50"/>
+  <pattern id="p2">
+    <pattern id="p1">

[f(malita), 2016/03/02 18:24:50]

One way to look at solving the problem within the current framework is that p2
should not attempt to layout p1 - it should only layout structural elements not
definition elements.  Definition elements get laid out on reference, so it
should be safe to skip at this point, no?

[fs, 2016/03/02 19:13:24]

It can't be skipped always though I think (i.e when there are no references to
#p1), or we might end up with a dirty tree in the end?

+      <rect fill="url(#p2)" x="10" y="10" width="200" height="200"/>
+    </pattern>
+  </pattern>
+</svg>