Support Chat completions (with custom LLM)

[murdercode]

As subject, we need to make the base uri configurable in env (fallback to the new /chat/completions).

This can be a break change, so goes up to v2

• [x] New endpoint (/chat/completions) as standard
• [x] Set gpt-3.5-turbo as default
• [x] Make the base uri configurable (for custom LLMs)

[murdercode]

Done https://github.com/The-3Labs-Team/tinymce-chatgpt-plugin/releases/tag/2.0.0