ohos: Implement voice() and setVoice()

voice() returned a default-constructed QVoice and setVoice() always
failed, so two tst_qtexttospeech cases failed on a device: voice(ohos)
requires a Ready engine to report a valid default voice, and
locale(ohos) requires that voice to be part of availableVoices().

Track the current voice instead of the current locale, and derive
locale() from it. availableVoices() keeps filtering by that locale, so a
voice reported by voice() is always one of them. setLocale() moves to
the first voice of the requested locale, and setVoice() accepts any
installed voice, adopting its locale.

Task-number: QTBUG-149561
Pick-to: 6.12
Change-Id: I75b7403b3f7c2240c2c54344fc30014be3ff6120
Reviewed-by: Dawid Śliwa <dawid.sliwa@qt.io>
Reviewed-by: Liang Qi <liang.qi@qt.io>