Delete FunctionGroupBase from Pepper.

Delete FunctionGroupBase from Pepper.

The FunctionGroupBase stuff was intended to used for our manual RTTI for
interfaces not associated with resources. However, we've been putting most
of those interfaces on the PPB_Instance_API which allows us to skip a
whole lot of routing, setup, and boilerplate code.

This patch moves the two remaining classes deriving from FunctionGroupBase
and moves them to special getters on the globals class. We'll keep these
classes around and since there are only two, it seems to warrant the special
case.

BUG=
TEST=

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=134509

[brettw, 2012-04-21 19:50:01]

Sorry about the big change, it's mostly mechanical and I already have reviews
out for almost everybody.

This patch depends on 
http://codereview.chromium.org/10170014/
which depends on
http://codereview.chromium.org/10163012/
which depends on
http://codereview.chromium.org/10091003/
none of which are checked in yet (in case you're comparing the code on trunk).

https://chromiumcodereview.appspot.com/10168026/diff/1/ppapi/proxy/ppb_instan...
File ppapi/proxy/ppb_instance_proxy.cc (right):

https://chromiumcodereview.appspot.com/10168026/diff/1/ppapi/proxy/ppb_instan...
ppapi/proxy/ppb_instance_proxy.cc:629: pp::CompletionCallback cb =
callback_factory_.NewCallback(
This changed because I deleted EnterHostFunctionForceCallback since this was the
only place that class was used and I don't expect other cases (since I don't
think we should have callbacks without a resource).

https://chromiumcodereview.appspot.com/10168026/diff/1/ppapi/thunk/enter.h
File ppapi/thunk/enter.h (right):

https://chromiumcodereview.appspot.com/10168026/diff/1/ppapi/thunk/enter.h#ne...
ppapi/thunk/enter.h:188: class PPAPI_THUNK_EXPORT EnterInstanceNoLock
I basically copied the code for the regular and "NoLock" case. doing this with
inheritance or templates is awkward and would require inlining these classes.
Since each one requires so little code and is used in so many places, it seems
better to duplicate that one line in the constructor.

https://chromiumcodereview.appspot.com/10168026/diff/1/ppapi/thunk/ppb_direct...
File ppapi/thunk/ppb_directory_reader_thunk.cc (left):

https://chromiumcodereview.appspot.com/10168026/diff/1/ppapi/thunk/ppb_direct...
ppapi/thunk/ppb_directory_reader_thunk.cc:19:
EnterFunctionGivenResource<ResourceCreationAPI> enter(directory_ref, true);
I also deleted EnterFunctionGivenResource since it's only used in 2 places, and
just hardcoded the lookup.

https://chromiumcodereview.appspot.com/10168026/diff/1/webkit/plugins/ppapi/h...
File webkit/plugins/ppapi/host_globals.cc (left):

https://chromiumcodereview.appspot.com/10168026/diff/1/webkit/plugins/ppapi/h...
webkit/plugins/ppapi/host_globals.cc:77: struct HostGlobals::InstanceData {
I deleted this since I deleted the function_proxies array so I could just store
the PluginInstance var directly in the map.

[jschuh, 2012-04-23 17:23:49]

Is this revenge for all those sandbox_policy rubberstamps? Because, I don't know
what most of this is doing.

[jschuh, 2012-04-23 19:04:23]

lgtm - I don't understand what this is doing beyond the mechanical part.