Server: handle generation until context is filled

[VJHack]

• I have read the contributing guidelines
• Self-reported review complexity:
  • Low
  • Medium
  • High

This PR fixes bug #9933. When n_predict is set to -2, generation will occur until the context is filled. Some basic server tests were added to demonstrate functionality.

[ngxson]

I don't want to discourage new people to make contributions to the project. But to be honest, coding style in this PR does not even meet the minimum requirement of this project. You should probably take more time to read the code.

[VJHack]

I don't want to discourage new people to make contributions to the project. But to be honest, coding style in this PR does not even meet the minimum requirement of this project. You should probably take more time to read the code.

I appreciate you taking the time to review my work. I'm new to this project and am still working on my coding style. Your feedback definitely helped a lot. I'll be more careful in my future contributions. Thank you!