Make classes and structures in core/events and core/fetch fast-allocated.

Source/core/events/EventTarget.h

Index: Source/core/events/EventTarget.h
diff --git a/Source/core/events/EventTarget.h b/Source/core/events/EventTarget.h
index 461ed23188d39d2ff9d3956db3b7b1f8de286cc9..1affa472120c06dad5662e9987977f8d16026572 100644
--- a/Source/core/events/EventTarget.h
+++ b/Source/core/events/EventTarget.h
@@ -39,6 +39,7 @@
 #include "core/EventTypeNames.h"
 #include "core/events/EventListenerMap.h"
 #include "platform/heap/Handle.h"
+#include "wtf/Allocator.h"
 #include "wtf/text/AtomicString.h"
 
 namespace blink {
@@ -50,6 +51,7 @@ class MessagePort;
 class Node;
 
 struct FiringEventIterator {
+    ALLOW_ONLY_INLINE_ALLOCATION();

[Yuta Kitamura, 2015/10/26 08:15:54]

Is this needed?

FiringEventIterator only contains off-heap objects, so having
ALLOW_ONLY_INLINE_ALLOCATION() here seems meritless.

[tasak, 2015/10/26 08:19:01]

I added this to ensure that no one invokes FiringEventIterator's new / delete.

     FiringEventIterator(const AtomicString& eventType, size_t& iterator, size_t& end)
         : eventType(eventType)
         , iterator(iterator)