Use keyed-call inline caches in delegating yield

Use keyed-call inline caches in delegating yield

Since we can't assume anything about the shape of the iterator in a
yield* (delegating yield), use an IC to do the next() and throw()
iterator method calls.

BUG=v8:2691
R=rossberg@chromium.org
TEST=mjsunit/regress/regress-2691

Committed: http://code.google.com/p/v8/source/detail?r=15111

[wingo, 2013-06-07 12:51:43]

Ping on this one. I don't think there is any ordering issue between this and
anything else.  Renaming send to next can happen in another commit.

[wingo, 2013-06-10 14:58:21]

Rebased onto master and adapted for the send -> next renaming.  Would be nice to
get this one in before the branch.

[wingo, 2013-06-12 13:13:07]

Adding Andreas as a possible reviewer :-)

This fixes http://code.google.com/p/v8/issues/detail?id=2691.

[rossberg, 2013-06-13 09:34:22]

lgtm

[wingo, 2013-06-13 10:18:38]

Committed patchset #3 manually as r15111 (presubmit successful).