Correctly materialize arguments object in Runtime_DebugEvaluate.

Correctly materialize arguments object in Runtime_DebugEvaluate.

The problem was that if the # arguments specified in the function
declaration and the # arguments passed to the function are not
the same, we use an arguments adapter frame to make it work. This
confuses the existing implementation to materialize the arguments
object.

R=peter.rybin@gmail.com
BUG=222893
Committed: https://code.google.com/p/v8/source/detail?r=14059

[Peter Rybin, 2013-03-23 04:17:41]

LGTM

I also have couple of comments about apply

1. My patch utility failed to apply this patch correctly, as it thought test
file was renamed, not new. chromiumcodereview.appspot.com site does similar
mistake. Probably it's a usual behavior, just wanted to warn.

2. Is calling "apply" on function enough for testing? Is it the same as calling
function with actual number of parameters? Just asking.

[Yang, 2013-03-25 10:01:11]

On 2013/03/23 04:17:41, Peter Rybin wrote:
> LGTM
> 
> I also have couple of comments about apply
> 
> 1. My patch utility failed to apply this patch correctly, as it thought test
> file was renamed, not new. http://chromiumcodereview.appspot.com site does
similar
> mistake. Probably it's a usual behavior, just wanted to warn.
> 
> 2. Is calling "apply" on function enough for testing? Is it the same as
calling
> function with actual number of parameters? Just asking.

1. I've encountered this before when I landed a patch for somebody else, but in
this case it should be fine, as I have the original git branch.
2. I verified that function apply works. But nevertheless I also added the
non-apply version just to be extra sure.

[Yang, 2013-03-25 10:02:03]

Committed patchset #1 manually as r14059 (presubmit successful).