feat: limit transcription output length based on input

Using heuristics. Also adds documentation and initial unit tests.

ref: N25B-209

src/control_backend/agents/transcription/__init__.py

@@ -1 +1,2 @@
+from .speech_recognizer import SpeechRecognizer as SpeechRecognizer
 from .transcription_agent import TranscriptionAgent as TranscriptionAgent