PlugValueWidgetTest : Fix handling of background updates

[johnhaddon]

This fixes test failures that looked like this :

Traceback (most recent call last):
  File "D:\a\gaffer\gaffer\build\python\GafferUITest\PlugValueWidgetTest.py", line 453, in testContextTrackerUpdates
    self.waitForUpdate( widget )
  File "D:\a\gaffer\gaffer\build\python\GafferUITest\PlugValueWidgetTest.py", line 62, in waitForUpdate
    handler.assertCalled()
  File "D:\a\gaffer\gaffer\build\python\GafferTest\ParallelAlgoTest.py", line 92, in assertCalled
    self.receive( timeout )()
    ^^^^^^^^^^^^^^^^^^^^^^^
  File "D:\a\gaffer\gaffer\build\python\GafferTest\ParallelAlgoTest.py", line 86, in receive
    raise AssertionError( "UIThread call not made within {} seconds".format( timeout ) )
AssertionError: UIThread call not made within 30.0 seconds

Our old approach of calling waitForUpdate() after running the code to trigger the update was bogus - it was possible for the update to have finished before we even got to calling waitForUpdate(). By triggering the update from inside the UIThreadCallHandler's scope we avoid this possibility.

[johnhaddon]

Looks like I failed to notice that other tests were using PlugValueWidgetTest.waitForUpdate(). I'll fix those tomorrow.

[johnhaddon]

I've pushed a new version which should take care of the other spots that use waitForUpdate(). It's a tiny bit annoying in that a lot of code gets indented one level further. A possible alternative to that would be for TestCase.setup() to install a UIThreadCallHandler that is removed by TestCase.tearDown(), so we always have one scoped and can drop the indentation. This would be a bigger change, but might be worth it.

It uses a new central GafferUITest.TestCase.uiThreadCallHandler that lets us avoid the race condition, and also makes it easier to handle UI thread calls in general.