[Accessibility] TTS API interface description + reqs update

[Jonathan Duddington]

message_id defer(void);
say_deferred(message_id, position, position_type);

If an application defers a message during speaking, then to resume, it
needs to specify a position from which to resume the speaking.  There
is no way to specify "continue from where it was deferred", or "from
the start of the sentence which was speaking".

This implies that defer() and say_deferred() are only applicable when
the application has chosen the AUDIO_OUTPUT_RETRIEVE method rather than
the AUDIO_OUTPUT_PLAYBACK method, because only then does the
application know where in the message the defer occurred.  Perhaps that
should be stated explicitly if that's the intention.

--