TEST CODE - Replace frame/iframe cache sets with boolean

TEST CODE - Replace frame/iframe cache sets with boolean

Remove 'CPluginDomTraverserBase::m_documentName', which was never explicitly
initialized. It was effectively a constant whose value was the empty string.

Replace set 'm_cacheDocumentHasIframes' with boolean 'cacheHasIframes' and
likewise for frames. The only value ever inserted into these sets was
'm_documentName', that is, the empty string, so these sets only existed in one
of two states: (1) empty (replaced by the value 'false'), and (2) a set
containing exactly one element, the empty string (replaced by the value
'true'). Initialize the booleans to true upon construction. Replace the
'insert()' operation with an assignment of 'true'. Replace the 'clear()'
operation with an assignment of 'false'. Replace the 'find()' operation with
the value of the replacement boolean value.

--------

Depends upon the following review:
    https://codereview.adblockplus.org/29332057/

[Eric, 2015-12-07 17:36:13]

I've made the original subject "TEST CODE" because I am slightly incredulous
that this actually works. I think we can commit this code, but I can't shake the
feeling that I'm missing something.

I came across the variable 'm_documentName' while doing 'CString' refactoring. A
simple "find all references" to see what it was triggered the present review.
Oleksandr, do you know if this member variable used in the past for anything?

[Oleksandr, 2015-12-14 10:09:09]

We shouldn't remove the m_documentName. The fact that it wasn't use is a bug,
but there already a proposed fix here:
https://codereview.adblockplus.org/29331669/. It is used for the Matches call
from the traverser.

[Eric, 2015-12-14 13:44:13]

On 2015/12/14 10:09:09, Oleksandr wrote:
> We shouldn't remove the m_documentName. The fact that it wasn't use is a bug,
> but there already a proposed fix here:
> https://codereview.adblockplus.org/29331669/. It is used for the Matches call
> from the traverser.

OK.

I have to say, though, that those cache sets still look kind of suspicious to
me. Let me repeat my suggestion to split up that review so we can commit the
documentUrl parts quickly. I think I'd like to revisit them once they might
actually do something. I didn't touch the synchronization around the cache sets
in this review, but it's not clear to me that they're necessary.

Closing this review as "defect corrected in another way".