Change how PepperMessageFilter is constructed, to simplify its use.

Change how PepperMessageFilter is constructed, to simplify its use.
Changes the ctor's HostResolver parameter to render_process_id, and
adds code to get the HostResolver. Modifies the single call site, and
delegate objects that actually call the constructor.
BUG=116317
TEST=existing, no new functionality.

[bbudge, 2012-09-07 23:56:49]

I think I was able to work around the PepperTCPSocket problem by passing the
HostResolver in the Connect method, rather than having it call back to the PMF.

Will the trybots test this to make sure I don't break Flash / Flapper?

http://codereview.chromium.org/10919169/diff/9002/content/browser/renderer_ho...
File content/browser/renderer_host/pepper/pepper_message_filter.cc (right):

http://codereview.chromium.org/10919169/diff/9002/content/browser/renderer_ho...
content/browser/renderer_host/pepper/pepper_message_filter.cc:732: if
(resource_context_) {
I'm not sure if we can change the thread that this is called on. If we can
switch it to the UI thread, we could get rid of the first PMF constructor
(RENDERER case) and just have one constructor taking the render_process_id.

[brettw, 2012-09-10 22:27:11]

http://codereview.chromium.org/10919169/diff/9002/content/browser/plugin_serv...
File content/browser/plugin_service_impl.h (right):

http://codereview.chromium.org/10919169/diff/9002/content/browser/plugin_serv...
content/browser/plugin_service_impl.h:131: int render_process_id,
Can you add a comment about what this is?

http://codereview.chromium.org/10919169/diff/9002/content/browser/renderer_ho...
File content/browser/renderer_host/pepper/pepper_message_filter.cc (right):

http://codereview.chromium.org/10919169/diff/9002/content/browser/renderer_ho...
content/browser/renderer_host/pepper/pepper_message_filter.cc:91:
process_id_(render_process_id),
I'm not totally following this. In the case of Flash, there will be one Flash
process and one PpapiPluginProcessHost in the browser. There will be many
renderers talking to it. The reason this process ID was previously set to 0 is
that there is no "the renderer" that we're talking to, we're talking to many
renderers in the case where type == PLUGIN.

Do we need a new type or something?