Add caching of the speech rate and language in android.speech.tts.TextToSpeech

so the speech params can be passed along in calls for text synthesis.