ccameron's https://codereview.chromium.org/23191007 with some changes to work

ccameron's https://codereview.chromium.org/23191007 with some changes to work
with https://codereview.chromium.org/23537016/ and
https://codereview.chromium.org/23880006/

BUG=

[sadrul, 2013-09-05 14:36:07]

This seems to build+run OK for me.

The fist CL in this series: https://codereview.chromium.org/23537016/
the second: https://codereview.chromium.org/23880006/
and this is the last one.

All three combined are in patchset3 here
(https://codereview.chromium.org/23450020/#ps10001)

Although it seems to work OK for me, can you give it a try and see if this
actually works as you expect? If so, I can start cleaning up the first two
patches and start pestering owners about them.

[ccameron, 2013-09-05 16:51:37]

Thanks!!

These and the others lg for me. I can take over the gl_surface_glx parts of this
(cause they need some work), but the rest can go in as-is (well, OWNER reviews
and all that).

It looks like this sequence is doing the reverse of the change I was making --
instead of having the GTK observers derive from a different (from
MessageLoop::Observer) type, we're having the NativeViewXExposeEventForwarder
class derive from a non-MessageLoop::Observer type. Seems perfectly reasonable
to me.