[feature] Lite-whisper family models support

[pzoltowski]

Would be great to have support for recent Lite Whisper series of models.

They have recently been incorporated into transformers.js with models in ONNX format here:

https://huggingface.co/onnx-community/lite-whisper-large-v3-turbo-fast-ONNX https://huggingface.co/onnx-community/lite-whisper-large-v3-turbo-acc-ONNX https://huggingface.co/onnx-community/lite-whisper-large-v3-turbo-ONNX

and ported from original pytorch models here:

https://github.com/efeslab/LiteASR https://huggingface.co/efficient-speech/lite-whisper-large-v3-turbo-acc

Paper (https://arxiv.org/abs/2502.20583) shows ~1.3x to ~1.6x speedup factor comparing to original openai models