Fixed Compiler Error C2864 - only static const integral data members can be initialized within a class

Vastly improved the joystick documentation, removed some redundant/unnecessary code, and fixed a couple of small bugs. This code should be ready for live testing.

Have settings loaded/stored correctly for all joysticks based on their UAS type and autopilot type. Switching between UASes doesn't work right yet, however.

The default UAS.cc now lists available actions that it can be commanded. These are polled by the Joystick Configuration window for each joystick button so that they can be assigned to trigger this action. Button presses don't yet trigger these actions, nor are they stored and reloaded on UAS switch.

The joystick code now understands if a UAS can reverse or not (based on UASInterface::systemCanReverse) and scales the joystick input accordingly. This isn't ideal on auto-centering axes and so the limit-range code should be brought back as an option for the throttle channel.

The UASInterface now defines a systemCanReverse() function for indicating if the UAS is capable of commanding a reverse thrust.

Removing UASes now returns the UI to an almost pristine state. This was done by altering the activeUASSet signal so that it can emit a NULL UAS. Crashes may still exist in autopilot-specific code.

Partway through refactoring of UAS deletion. This is in order to actually support removing UASes from QGC, which is currently impossible.

Removing UASes now returns the UI to an almost pristine state. This was done by altering the activeUASSet signal so that it can emit a NULL UAS. Crashes may still exist in autopilot-specific code.

Partway through refactoring of UAS deletion. This is in order to actually support removing UASes from QGC, which is currently impossible.

Impl. Lorenz suggestions wrt one velocityChanged, altitudeChanged, speedChanged signal per reference frame. Impl. CDI on compass disk.

Fix for static const float cannot be initialized within a class definition, and other windows compile errors.

Hotfix: Fixed widget / UAS initialization order which led to text messages (and other updates) being ignored if the widget was not open when the UAS was created

Changes for new calibration in APM 3.0 (Re: https://github.com/diydrones/MissionPlanner/issues/38), and change in some of the serial timeouts