whisper-medium q8 model

Xenova/whisper-medium ONNX q8 quantized weights for the dsh-voice plugin. Extracts to a model directory with an onnx/ subfolder (config.json, tokenizer.json, onnx/encoder_model_quantized.onnx, onnx/decoder_model_merged_quantized.onnx, ...).