v0.4.0

Added

• Chunking options to speech-to-text to support long audio transcription (#236)
• Support for Whisper timestamps and task/language configuration (#238)
• Support for streaming speech-to-text results (#242)
• Introduced featurizer batch phase that is compiled as part of the serving computation (#243)
• Removed possibly contentious Nx calls from serving postprocessing (#244, #245)

Changed

• Deprecated Bumblebee.Audio.speech_to_text/5 in favour of the more specific Bumblebee.Audio.speech_to_text_whisper/5
• Changed the tensors returned from embedding servings to use Nx.BinaryBackend