Refactor tool-speechtotext: extract sttlib shared library and add tests

Extract duplicated code (Whisper loading, audio recording, transcription,
VAD processing) into reusable sttlib/ package. Rewrite all 3 scripts as
thin wrappers. Add 24 unit tests with mocked hardware. Fix GPU fallback
bug in assistant.py and args.system assignment bug.

python/tool-speechtotext/sttlib/__init__.py

@@ -0,0 +1,7 @@
+from sttlib.whisper_loader import load_whisper_model
+from sttlib.audio import record_until_enter, pcm_bytes_to_float32
+from sttlib.transcription import transcribe, is_hallucination, HALLUCINATION_PATTERNS
+from sttlib.vad import (
+    VADProcessor, audio_callback, audio_queue,
+    SAMPLE_RATE, CHANNELS, FRAME_DURATION_MS, FRAME_SIZE, MIN_UTTERANCE_FRAMES,
+)