Fix adopting a video element with controls

Source/core/html/HTMLMediaElement.cpp

Index: Source/core/html/HTMLMediaElement.cpp
diff --git a/Source/core/html/HTMLMediaElement.cpp b/Source/core/html/HTMLMediaElement.cpp
index a709141682132ddc3829fd251ecc4500baed28bc..c829246ade5e89bb6c32183ee8e2cc033eff00ea 100644
--- a/Source/core/html/HTMLMediaElement.cpp
+++ b/Source/core/html/HTMLMediaElement.cpp
@@ -827,12 +827,11 @@ void HTMLMediaElement::prepareForLoad()
     // algorithm, but do it now because we won't start that until after the timer fires and the
     // event may have already fired by then.
     setShouldDelayLoadEvent(true);
-
-    configureMediaControls();
 }
 
 void HTMLMediaElement::loadInternal()
 {
+    configureMediaControls();

[philipj_slow, 2014/09/16 11:59:39]

All existing tests and the one you've added pass if one just removes
configureMediaControls() above but does not add this one, i.e. this line is not
really tested.

A smaller fix is to replace configureMediaControls() with if
(hasMediaControls()) mediaControls()->reset(). AFAICT, the purpose of
configureMediaControls() in prepareForLoad() is to update the controls since the
currentTime, duration and other things may have changed, and if there are no
controls there's no point in creating them at this point I believe. When moving
to a new document, HTMLMediaElement::didNotifySubtreeInsertionsToDocument() is
going to call configureMediaControls() and otherwise parseAttribute() should do
it.

I suggest the more conservative fix, along with a test that would fail if the
code were removed entirely, to prove that it's actually needed there.

[aberent, 2014/09/16 14:45:40]

Having failed to create a test in a bit more than your suggested 30 minutes I
will go for the conservative fix without the test.

     // HTMLMediaElement::textTracksAreReady will need "... the text tracks whose mode was not in the
     // disabled state when the element's resource selection algorithm last started".
     m_textTracksWhenResourceSelectionBegan.clear();