feat: Jan supports most llama.cpp params

[imtuyethan]

Goal

• Jan supports most llama.cpp params

Tasklist

Cortex

• [x] https://github.com/janhq/cortex.cpp/issues/1151

Jan

• [ ] Update Right Sidebar UX for Jan
• [ ] Enable Jan's API server to pass through most llama.cpp params

Related

• https://github.com/janhq/jan/issues/3140
• Beam search? https://github.com/janhq/jan/issues/3112

Original Post

- [X] I have searched the existing issues

### Is your feature request related to a problem? Please describe it

- llama.cpp Settings (e.g. attention) should be consistent across llama.cpp, Cortex and Jan
- From an Eng perspective, we should ensure llama.cpp settings get bubbled up to Cortex and Jan

### Describe the solution

- [ ] Identify all relevant model settings that need to be synced
- [ ] Design a common format for representing these settings across all projects
- [ ] Jan Model Settings should follow common format
- [ ] Cortex should allow user to pass inference-time and runtime parameters
- [ ] Process for llama.cpp updates (who should this be driven by?)

### Teachability, documentation, adoption, migration strategy

-

### What is the motivation / use case for changing the behavior?

[0xSage]

Related #3140