feat: Implement TensorRT-LLM Extension

[imtuyethan]

Specs

https://www.notion.so/jan-ai/TensorRT-LLM-C-Inference-Extension-for-Jan-Desktop-Server-32b5101c595a49489cfe37680e769dd2?pvs=4

Design

Figma: https://www.figma.com/file/ytn1nRZ17FUmJHTlhmZB9f/Jan-App-(version-1)?type=design&node-id=5178-216651&mode=design&t=Ld1o

ytKp1oD4ElKt-4

[Van-QA]

@louis-jan, please help us resolve these issues:

1. The extension still appears in Jan Windows on a PC with no GPU ❌ clicking on install doesn't install anything + no toast message / warning to users, nothing happen.
   

2. Install UI on dark mode:
   

3. Missing label on the new model:
   

4. Unclear error toast title ❌ only "Failed"
   

[Van-QA]

5. Missing recommendation label ❌
   

6. Need consistency in the TensorRT label ❌
   

[namchuai]

issue 5&6 is addressed in the PR #2346

[Van-QA]

7. TensorRT model does not functioning with RAG or with Stream off ❌
   

[Van-QA]

8. Swagger chat completion is incompatible with TensorRT model for the local API server ❌

9. On Windows with no GPU, the incompatible value is incorrect ❌
   

10. Extra space in the extension name:
    

[Van-QA]

Tested and looking good using Jan v0.4.8-333 ✅