fix a typo in detect_language_multi_segment comment

[fcakyon]

Hello,

There is a typo in a parameter explanation. This definition is incorrect:

faster-whisper/faster_whisper/transcribe.py

Line 1863 in d57c5b4

# The default is 0.02 (2.0%) i.e, if more than 2.0% of the audio is silent,

This definition is the correct one:

faster-whisper/faster_whisper/transcribe.py

Line 1917 in d57c5b4

# if the audio after VAD is less than 2% of the original audio, consider it as silence

I updated the incorrect comment to reflect the correct meaning.

cc: @MahmoudAshraf97