Request: Update LlamaLib to Support Gemma3n (like in Issue #323)

[TarekMarrawi]

Describe the feature

I’d like to request an official update of LlamaLib to use a newer llama.cpp version that supports Gemma3n, similar to what was done in #323.

Context
I attempted to manually replace llama.cpp with the latest from ggerganov/llama.cpp at commit 8846aace, which adds Gemma3n support.

However, the build fails due to:

Missing server.cpp referenced in undreamai.h

Signature mismatch in handle_slots_action(json) vs handle_slots_action(json, httplib::Response*)

Missing json.hpp

My Goal
Enable Gemma3n inside Unity via LLMUnity (e.g., llmunity_gemma3n backend), using updated LlamaLib.

Feature Request
Could you please:

Update the llama.cpp submodule in LlamaLib to support Gemma3n

Patch undreamai.cpp to match the new llama.cpp APIs

Ensure missing files (like server.cpp) are either committed or removed

Happy to help test this if there's a working branch or PR. Thanks for the amazing work so far.

[amakropoulos]

Thank you for the feature request!
I'm currently rewriting LlamaLib (the LLM backend) and will include the latest updates from llama.cpp.
I'm expecting to integrate it in LLMUnity in the next 1-2 weeks.