Link to issue number:

None

Summary of the issue:

Currently, system tests use pyautogui to send key presses to NvDA and the OS. However, a limitation of pyautogui is that it cannot send extended keys. For example, it is impossible to send a downArrow key press for the extended arrows (upside down t). It supports "down" but that is on the numberpad.
Also, the strings it uses for keys vary quite a bit from NvDA's identifiers.
There are system tests that require sending extended keys. The first being in pr #11569

Description of how this pull request fixes the issue:

Rather than sending keys with pyautogui, instead send them via NVDA's own input gesture framework.
Specifically:

• Add an emulateKeyPress method to the Robot spy object, which takes an NVDA kb identifier, and emulates the gesture. Using inputCore.manager.emulateGesture means that either a script will be executed, or the key will be sent on to the OS.
  Also, this new method by default blocks until the action has definitely been executed. It does this by first inserting a function into NVDA's event queue after emulating the gesture, and then waiting on that function to be executed. Finally, it then waits on NVDA's core to fall asleep (meaning that no more events or speech is being processed). This makes the tests much more robust as in some cases if a key press resulted in two or more speak calls to the synth, only the first would be waited for. Again, this is necessary for pr Announce entering a listItem when moving by character or word in contentEditables and browse mode  #11569

Testing performed:

All the system tests pass.

Known issues with pull request:

None.

Change log entry:

Changes for developers:
System tests can send keys using spy.emulateKeyPress, which takes a key identifier that conforms to NvDA's own key names.