This was crashing in CppunitTest_sc_macros_test on Windows with
--enable-mergelibs (release build), because the first invocation of
processWindowResizeEvent() deleted the vcl::Window, and the
maControllers.count(pWindow) test creates a VclPtr for it, so ends up
with a double-free.

TODO: is processWindowResizeEvent() supposed to be idempotent?
It would be possible to detect that there is already an event posted by
checking m_PostedWindows in postWindowResizeEvent().

Change-Id: I7b72f2baf21bb8223e9fe4bd929d826217b920e5
(cherry picked from commit 520514cf)
Reviewed-on: https://gerrit.libreoffice.org/19022Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>