v0.1.3-alpha Stability increase

• save cache directly after each default string was fetched
• add OnDone callback to SynthesisListener for that
• add option to trigger a save cache for that as well
• fix problems with the buffer size in Mimic3TTSEngineWeb while synthesizing text. Use TextToSpeech.getMaxSpeechInputLength() to determine the maximum buffer size supported
• only make audio available after the fetch is completed. May have introduced white noise before due to slow connection. (when the bytes come slower than the bitrate of the audio playback is)
• trim the text input when making the request to the web server