Refactor scope and function state tracking in (Pre)Parser.

Refactor scope and function state tracking in (Pre)Parser.

Notes:
- PreParser::Scope was a weird combination of Parser::FunctionState and
  Scope. Split it into two (PreParser::FunctionState and PreParser::Scope). This
  is necessary for unifying the Parser and the PreParser.
- Scopes take care of language mode and tracking "with".
- FunctionStates take care of counting material literal indexes, properties
  etc. and tracking generators.
- PreParser::Scope::InsideWith was a hack to make a FunctionState-like object
  take care of tracking "with". It's now the responsibility fo PreParser::Scope
  and Scope.
- PreParser::ScopeType is unnecessarly, there is already a ScopeType enum in
v8globals.h.
- Renamed scope stack variables so that they're consistent in Parser and PreParser.
- Parser::FunctionState and Parser::BlockState had an unnecessary dependency to
  the Parser; they only need a couple of things from Parser. Broke the
  dependency.


R=ulan@chromium.org
BUG=v8:3126
LOG=N

Committed: https://code.google.com/p/v8/source/detail?r=19319

[ulan, 2014-02-12 10:16:50]

lgtm

https://codereview.chromium.org/148293011/diff/60001/src/parser.cc
File src/parser.cc (right):

https://codereview.chromium.org/148293011/diff/60001/src/parser.cc#newcode498
src/parser.cc:498: outer_scope_(*scope_stack),
We have two many scope stacks =)
Does this precondition hold: *scope_stack == scope->outer_scope_?
If yes could you assert this or get rid of outer_scope_ here.

[marja, 2014-02-12 11:36:56]

Thanks for review! The simplification doesn't unfortunately work :(

https://codereview.chromium.org/148293011/diff/60001/src/parser.cc
File src/parser.cc (right):

https://codereview.chromium.org/148293011/diff/60001/src/parser.cc#newcode498
src/parser.cc:498: outer_scope_(*scope_stack),
On 2014/02/12 10:16:51, ulan wrote:
> We have two many scope stacks =)
> Does this precondition hold: *scope_stack == scope->outer_scope_?
> If yes could you assert this or get rid of outer_scope_ here.

It doesn't hold. In Parser::ParseLazy we seem to push a scope to the scope stack
which has outer_scope == NULL, even though there are some scopes in the stack
already. Scope::DeserializeScopeChain seems to complicate things.

-> outer_scope_ was here for a reason, leaving it like this.

[marja, 2014-02-12 12:02:14]

Committed patchset #5 manually as r19319 (presubmit successful).