Possible fix for a Mac performance regression.

Possible fix for a Mac performance regression.

Introduced in
http://src.chromium.org/viewvc/blink/trunk/Source/wtf/Vector.h?revision=157005

We now do the minimal amount of work possible on Vector destruction. We should
now be faster than ever before:
- The most common case is actually a totally unused Vector with no inline
storage. In this case, we can bail and do no other work, if the buffer pointer
is NULL.
- There's no need to set m_size to 0 in the destructor; we've already checked
that it is 0 and if not called resize(0) which will set it to 0.
- In the Vector with inline storage case, we once again check m_size before
calling shrink(), which might not be inline. It is common for m_size to be 0.

This also saves >50KB on a GCC Release build. clang, as usual, is barely
affected on account of its less aggressive inlining.


BUG=283760
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=157198

[Tom Sepez, 2013-09-04 01:13:22]

lgtm

[commit-bot: I haz the power, 2013-09-04 02:49:44]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/cevans@chromium.org/23455034/1

[commit-bot: I haz the power, 2013-09-04 05:13:00]

Change committed as 157198