fix: reverted the URL of llama.cpp back to 'completion'.

[ipaddicting]

Description

Closes: #5530
Reverted the URL of llama.cpp back to 'completion' since it's not changed at all.

The latest documentation from llama.cpp:
https://github.com/ggml-org/llama.cpp/blob/master/tools/server/README.md#post-completion-given-a-prompt-it-returns-the-predicted-completion

/v1/completions is for OAI-compatible clients only, but in LlamaCpp.ts there is no /v1 presented, and I don't think it should.

Tests

Local tests with the extension built.

[netlify]

✅ Deploy Preview for continuedev canceled.

Name	Link
🔨 Latest commit	e8523a2
🔍 Latest deploy log	https://app.netlify.com/projects/continuedev/deploys/682af21731c9e800089d265a

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[ipaddicting]

I have read the CLA Document and I hereby sign the CLA

[tomasz-stefaniak]

Thanks!

For some extra background, I think we may actually want to remove this provider in the future, given that llama.cpp is now OpenAI-compatible. That means the OpenAI provider can be used instead.