AdbMouse and AdbKeyboard are subdevices of the CUDA device alongside AdbBus.
This doesn't make sense because conceptually, ADB devices hang off of the ADB
bus, not CUDA itself. An ADB bus can exist without a CUDA present, for example
Egret on older 68K Macs and the PMU on newer Power Macs. Therefore, make the ADB
device list a subhierarchy of AdbBus instead. Add a new "adb_devices" property
belonging to AdbBus that can allow users to specify ADB devices on the command
line at machine creation time, independent of the emulated bus's host. Make this
property default to "Mouse,Keyboard" to preserve existing behavior.
Besides generating KeyboardEvents in the SDL event handler and
returning the key state in the register 0 reads of the AdbKeyboard
device, we also needed to generalize the ADB bus polling a bit. We now
check all devices that have the service request bit set, instead of
hardcoding the mouse.
The SDL key event -> ADB raw key code mapping is based on BasiliskII/
SheepShaver's, but cleaned up a bit.
Result of running IWYU (https://include-what-you-use.org/) and
applying most of the suggestions about unncessary includes and
forward declarations.
Was motivated by observing that <thread> was being included in
ppcopcodes.cpp even though it was unused (found while researching
the use of threads), but seems generally good to help with build
times and correctness.
