chore(whisper-cpp): Convert to Purego and add VAD

[richiejp]

Description

Converts the Whisper backend to use Purego similar to stablediffusion. Also adds some features which are not in the upstream CGO bindings.

Notes for Reviewers

• We could upstream the Purego bindings, but I'm not sure what that would look like, so will just try it here first.
• Initially I've added just a new VAD backend for testing, then will convert the rest.

Signed commits

• Yes, I signed my commits.

TODO:

• fix VAD end time (speech segments are detected, but RT API is not submitting for transcription after period of silence, possibly time units on segments are wrong)
• convert rest of whisper backend to purego
• fix transcription failed bug
• use transcriptions in-built VAD mode

[netlify]

✅ Deploy Preview for localai ready!

Name	Link
🔨 Latest commit	0345dfb
🔍 Latest deploy log	https://app.netlify.com/projects/localai/deploys/68a83eb0f179e30008a6da3b
😎 Deploy Preview	https://deploy-preview-6087--localai.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[richiejp]

ah now I realise that the VAD model can be combined with the transcribe model. So we can just call transcribe and it does VAD first and short circuits if no speech is detected. This changes a lot.