Add audio-to-audio pipeline tag

This PR adds the `audio-to-audio` pipeline tag to the model card metadata. This tag accurately reflects the model's primary function as an acoustic discrete codec tokenizer, converting audio to a discrete representation and back. This makes the model discoverable by users interested in audio-to-audio tasks.

README.md

@@ -1,5 +1,6 @@
 ---
 license: mit
+pipeline_tag: audio-to-audio
 ---
 
 # WavTokenizer