[cleanup] Update ObjectDataKind kind's comment

Bug: v8:7790
Change-Id: Ia3902c8f12e856a2879e58de496cdf6291267496
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2450199
Commit-Queue: Santiago Aboy Solanes <solanes@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70350}

src/compiler/js-heap-broker.cc

@@ -50,7 +50,7 @@ HEAP_BROKER_SERIALIZED_OBJECT_LIST(FORWARD_DECL)
 HEAP_BROKER_NEVER_SERIALIZED_OBJECT_LIST(FORWARD_DECL)
 #undef FORWARD_DECL
 
-// There are three kinds of ObjectData values.
+// There are five kinds of ObjectData values.
 //
 // kSmi: The underlying V8 object is a Smi and the data is an instance of the
 //   base class (ObjectData), i.e. it's basically just the handle.  Because the
@@ -65,6 +65,12 @@ HEAP_BROKER_NEVER_SERIALIZED_OBJECT_LIST(FORWARD_DECL)
 //   data is an instance of the base class (ObjectData), i.e. it basically
 //   carries no information other than the handle.
 //
+// kNeverSerializedHeapObject: The underlying V8 object is a (potentially
+//   mutable) HeapObject and the data is an instance of ObjectData. Its handle
+//   must be persistent so that the GC can update it at a safepoint. Via this
+//   handle, the object can be accessed concurrently to the main thread. To be
+//   used the flag --turbo-direct-heap-access must be on.
+//
 // kUnserializedReadOnlyHeapObject: The underlying V8 object is a read-only
 //   HeapObject and the data is an instance of ObjectData. For
 //   ReadOnlyHeapObjects, it is OK to access heap even from off-thread, so