[parser] PreParserExpression cleanup

[parser] PreParserExpression cleanup

Cleanup/refactoring CL:

- Previously dead code is no longer dead
- More information exposed to preparser about different literal kinds
  (should be useful for rewriting https://crrev.com/1168643005 in a
  simpler way)

BUG=
R=
LOG=N

[caitp (gmail), 2015-08-17 18:26:12]

Hi, can you take a look? Some refactoring/cleanup split apart from rewriting the
destructuring assignment CL. Cleans up some small inconsistencies and provides
some extra data to the preparser, which is very helpful for my destructuring
assignment rewrite.

https://codereview.chromium.org/1291343005/diff/1/src/preparser.h
File src/preparser.h (right):

https://codereview.chromium.org/1291343005/diff/1/src/preparser.h#newcode924
src/preparser.h:924: return
PreParserExpression(TypeField::encode(kStringLiteral));
the `Expression` suffix wasn't really adding anything

https://codereview.chromium.org/1291343005/diff/1/src/preparser.h#newcode938
src/preparser.h:938: return
PreParserExpression(TypeField::encode(kNumberLiteral));
Just here for consistency --- might be useful for enabling the preparser to
discern between smi-able numbers and float-only numbers, but that's speculating.

https://codereview.chromium.org/1291343005/diff/1/src/preparser.h#newcode942
src/preparser.h:942: return
PreParserExpression(TypeField::encode(kArrayLiteral));
I have uses for this in destructuring assignment, so that I can centralize a lot
of logic that was very distributed and convoluted in the old CL. Not positive if
it will work, but I'm going to try it on top of this CL first anyhow.

https://codereview.chromium.org/1291343005/diff/1/src/preparser.h#newcode1103
src/preparser.h:1103: typedef BitField<Type, 0, 4> TypeField;
I guess this doesn't really need the extra bit --- revert this line?

[adamk, 2015-08-17 18:40:17]

adamk@chromium.org changed reviewers:
+ adamk@chromium.org

[adamk, 2015-08-17 18:40:17]

A little confused by this patch...isn't this code basically still dead? I'd
rather see these changes as part of a patch that makes use of them (which it
sounds like is forthcoming, so maybe I'll just wait for that).

[caitp (gmail), 2015-08-17 18:44:22]

On 2015/08/17 18:40:17, adamk wrote:
> A little confused by this patch...isn't this code basically still dead? I'd
> rather see these changes as part of a patch that makes use of them (which it
> sounds like is forthcoming, so maybe I'll just wait for that).

It does add some dead code (the NumberLiteral stuff is dead even after the new
patch, so it's cool to remove that --- just there for consistency).
The Object/Array literal stuff is not dead after the forthcoming patch, and the
StringLiteral stuff was previously dead, but updated to be used (although it
admittedly isn't used in any useful way).

Anyways, other patch in progress, this one can wait on it