Revert of Oilpan: Remove raw pointer to DescendantInvalidationSet from StyleInvalidator

Revert of Oilpan: Remove raw pointer to DescendantInvalidationSet from StyleInvalidator::RecursionData (patchset #4 id:60001 of https://codereview.chromium.org/1225233005/ )

Reason for revert:
blink_perf.css regression
https://code.google.com/p/chromium/issues/detail?id=513378#makechanges

Original issue's description:
> Oilpan: Remove raw pointer to DescendantInvalidationSet from StyleInvalidator::RecursionData
> 
> BUG=509911
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=199288
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=199294

TBR=haraken@chromium.org,yutak@chromium.org,oilpan-reviews@chromium.org,sigbjornf@opera.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=509911
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=201869

[keishi, 2015-09-07 09:33:26]

Created Revert of Oilpan: Remove raw pointer to DescendantInvalidationSet from
StyleInvalidator::RecursionData

[haraken, 2015-09-07 09:35:08]

LGTM

Does this really affect performance?

[keishi, 2015-09-07 09:36:55]

On 2015/09/07 09:35:08, haraken wrote:
> LGTM
> 
> Does this really affect performance?

I doubt it but if the bisect points to this so I thought I'd revert it and see.

[haraken, 2015-09-07 09:44:04]

On 2015/09/07 09:36:55, keishi wrote:
> On 2015/09/07 09:35:08, haraken wrote:
> > LGTM
> > 
> > Does this really affect performance?
> 
> I doubt it but if the bisect points to this so I thought I'd revert it and
see.

Makes sense.