Revert: [k2-fsa#1521] No need to reset sample-buffer (k2-fsa#1524)

Co-authored-by: VEP <[email protected]>

python-api-examples/two-pass-speech-recognition-from-microphone.py

@@ -427,6 +427,7 @@ def main():
                     )
                     print("\r{}:{}".format(segment_id, result), flush=True)
                     segment_id += 1
+                else:
                     sample_buffers = []
 
                 first_recognizer.reset(stream)