Some changes in the tabbox for active

[Marco Martin]

Hi all,

in Plasma Active we are now using the qml tabbox as only task switcher. there 
were some issues tough, mostly have been resolved in the kwin/mart/tabboxFixes 
branch, some of them i think are potentially mergeable, some are quite ugly 
and more proper fixes would be needed probably.

now, the packages in active for kde-workspace have something like 14 patches 
and that's not soo nice..

of the most decent onles i'll do a review request.

* the main window of the active shell has the atom _KDE_FIRST_IN_WINDOWLIST to 
do what the name implies, be first in windowlist, but another thing is needed, 
not haveing the close button (and since is borderless i can't say it's not 
closeable apparently) https://projects.kde.org/projects/kde/kde-
workspace/repository/revisions/e5d4108b0cd8a12970cea809a65b6fb3c9363ba5

* when the tabbox is open and the same window is tapped again, re-activate it 
to close the tabbox (the window the tabbox is embedded into does have focus, 
but is not in the tabbox) https://projects.kde.org/projects/kde/kde-
workspace/repository/revisions/5d007717904ef922d50a9618d105e4c7c6651555

* as explained some days ago here, connecting to kwindowsystem signals in the 
tabbox seems to have catastrophic effects, using signals from TopLevel instead 
solves the issue: https://projects.kde.org/projects/kde/kde-
workspace/repository/revisions/d1a33ff24abf854c6463638411906718d4b06528 and 
https://projects.kde.org/projects/kde/kde-
workspace/repository/revisions/1d4038b39bf47bc6caefae0963b61c987d1ac759

* clientmodel.cpp, line 175,  if (partialReset && !m_clientList.isEmpty()) the 
check on partialResetmakes the window list appear empty sometimes: 
https://projects.kde.org/projects/kde/kde-
workspace/repository/revisions/e5d4108b0cd8a12970cea809a65b6fb3c9363ba5

* and now teh ugly: on meego and mer, it seems only xi2 events are delivered, 
in an x11Event() mouse press/release never arrive (i bet this is also causing 
some weird focus issues with the virtual keyboard we're having)
the windowstrip was translating and forwarding mouse events to the embedding 
window if any, to be able to drag it. mouseevents must be built starting from 
those xi2 events (i tried forwarding the xi2 events, but doesn't seem to work) 
https://projects.kde.org/projects/kde/kde-
workspace/repository/revisions/859e2f47c2a4052c3fd7e42cbc463bc9c9aa3a9b

Cheers,
Marco Martin