Move polymorphic stub computation and compilation to stub cache

Move polymorphic stub computation and compilation to stub cache

Committed: https://code.google.com/p/v8/source/detail?r=13478

[Toon Verwaest, 2013-01-22 17:49:58]

PTAL

In this CL I first moved polymorphic cache handling to stub-cache, so that it
matches other cache related code.

I made 2 copies of the code shared between KeyedStore and KeyedLoad, and
tailored it towards the particular use-cases (basically removed sharing of code;
which actually slightly shrunk the resulting code-size).

Finally I removed the KeyedIC superclass entirely, and changed the superclass of
KeyedStoreIC to StoreIC, and KeyedLoadIC to LoadIC. This allowed me to push the
Load and Store methods back down from IC to LoadIC and StoreIC respectively.

[Jakob Kummerow, 2013-01-23 13:55:06]

LGTM.

https://chromiumcodereview.appspot.com/11953025/diff/1005/src/ic.cc
File src/ic.cc (right):

https://chromiumcodereview.appspot.com/11953025/diff/1005/src/ic.cc#newcode1696
src/ic.cc:1696: case KeyedStoreIC::STORE_TRANSITION_SMI_TO_OBJECT:
nit: since this is now a method of KeyedStoreIC, you could also
s/KeyedStoreIC:://. I don't care much either way.

[Toon Verwaest, 2013-01-23 15:35:58]

Addressed comments,

https://chromiumcodereview.appspot.com/11953025/diff/1005/src/ic.cc
File src/ic.cc (right):

https://chromiumcodereview.appspot.com/11953025/diff/1005/src/ic.cc#newcode1696
src/ic.cc:1696: case KeyedStoreIC::STORE_TRANSITION_SMI_TO_OBJECT:
On 2013/01/23 13:55:06, Jakob wrote:
> nit: since this is now a method of KeyedStoreIC, you could also
> s/KeyedStoreIC:://. I don't care much either way.

Done.