Drop InvalidationSetMode.

Drop InvalidationSetMode.

*** NOT FOR LANDING ***

This is a demonstration of the end result of allowing all selectors to
update invalidation sets using updateInvalidationSets and never fall back
to a separate wholesubtree loop in collectFeaturesFromSelector. The method
for finding the InvalidationSetMode can be dropped completely, handling
selectors without features and selectors requiring subtree recalcs when
extracting the features instead.

By introducing a flag for crossing insertion points, the only special
handling left is to "clear" the InvalidationSetFeatures, and use a subtree
recalc when finding a ::first-line, ::first-letter, or :host-context in
the right-most compound selector.

There are a bunch of pseudo classes/elements which are white-listed in this
CL which needs to be white-listed on a case-by-case basis backed up with
tests, but there were no tests failing locally. I've added switch-asserts
causing compile errors if added selector types are neither white- nor
black-listed.

I have renamed the "wholeSubtree" flag to "adjacent" since it keeps track
of whether the last seen combinator was an adjacent combinator.

RuleFeaturesSet::collectFeaturesFromSelector is now back to how it was
before we introduced invalidation sets (without the feature hash sets).
Now, it only counts adjacent combinators and sets a flag when finding
::first-line, so it is possible to squash it into the same method as
updateInvalidationSets() to have a single pass over the selector when
finding rule features.

[rune, 2014-10-07 14:36:51]

rune@opera.com changed reviewers:
+ chrishtr@chromium.org, esprehn@chromium.org

[rune, 2014-10-07 14:36:51]

@chrishtr, @esprehn:

I'm trying to make invalidation set generation more straight-forward and
understandable with less special cases. If you like this approach, I'll go ahead
and do this with a sequence of smaller CLs.

[chrishtr, 2014-10-10 16:45:06]

Looks like a nice improvement.

https://codereview.chromium.org/639433002/diff/1/Source/core/css/RuleFeature.cpp
File Source/core/css/RuleFeature.cpp (right):

https://codereview.chromium.org/639433002/diff/1/Source/core/css/RuleFeature....
Source/core/css/RuleFeature.cpp:282: features.hasFeatures = false;
It's a weird to trigger subtree by setting hasFeatures = false. Took me a while
to figure out what you were doing here.

[rune, 2014-10-13 07:56:41]

On 2014/10/10 16:45:06, chrishtr wrote:
> Looks like a nice improvement.

Thanks! Unless @esprehn has objections about the direction, I'll continue
breaking this into smaller patches.

[rune, 2014-11-13 13:30:05]

Closed as this has now been replaced by smaller CLs. All but [1] landed.

[1] https://codereview.chromium.org/726503003