Merge lp://qastaging/~unity-2d-team/unity-2d/unity-2d_enable-super-numkey-combo into lp://qastaging/unity-2d

This is the heir of https://code.launchpad.net/~unity-2d-team/unity-2d/enable-super-numkey-combo/+merge/89883 that was rejected because was against unity-2d-shell instead of unity-2d

Copying here Michał's comments of what needs to be fixed.

The test is broken:
* if the launcher item is already in "running" state, the test won't really test anything
* if the app in question doesn't start within TIMEOUT, the test will fail even though the feature actually worked
* LibreOffice Impress, for example, doesn't go into "running" state before you actually create a slidedeck, so the test will fail in a clean environment every time
* the result of this test is a workspace cluttered with up to 10 new windows that aren't closed

1. I think that the logic based on isX11Keysym in LauncherView::forwardNumericHotkey(), as well as Hotkey constructor are good candidates for introducing two simple classes inherited from HotKey: NumericHotkey and NumpadNumHotKey that hide x11keysm-based logic. They would inherit from Hotkey and from a common abstract class that provides index() interface, hinding index = key - .... calculation/comparison logic. I don't insist, but I prefer to avoid leaking of logic outside of classes and use polymorphic classes instead.

2. I think HotkeyMonitor::getHotkeyFor(Qt::Key ...) should check for equality of isX11keysym flag, and not only key and modifiers values.

This superceded by https://code.launchpad.net/~stolowski/unity-2d/numpad-modifier-support/+merge/97931, rejecting