Expose GGUF audio support in gguf_lib

[Godzilla675]

Summary

• wire the new llama audio bridge through gguf_lib JNI/Kotlin APIs
• rebuild packaging so the shipped AAR works for both arm64-v8a and x86_64
• unblock ToolNeuron's end-to-end GGUF audio flow

Dependency

• Add GGUF audio bridge support llama.cpp-android#2

[Copilot]

Pull request overview

This PR extends the gguf_lib Android/JNI bridge to support VLM (vision/audio) projector loading and VLM streaming generation, and adds a runtime toggle for “thinking” blocks in chat templates. It also updates the native build to use a configurable llama.cpp-android checkout, compiles VLM engine sources, and enables x86_64 builds.

Changes:

• Add JNI + Kotlin APIs for VLM projector lifecycle (load/release/info/default marker) and VLM streaming generation with media payloads.
• Add a “thinking enabled” flag that is passed into chat template application.
• Update native build configuration (configurable LLAMA_DIR, VLM sources, and x86_64 ABI support with KleidiAI disabled on non-arm64).

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
gguf_lib/src/main/java/com/dark/gguf_lib/GGUFNativeLib.kt	Adds JNI declarations for thinking toggle + VLM projector and VLM generation APIs.
gguf_lib/src/main/java/com/dark/gguf_lib/GGMLEngine.kt	Exposes higher-level Kotlin wrappers for thinking toggle, VLM projector management, and a VLM Flow generator.
gguf_lib/src/main/cpp/gguf_lib.cpp	Implements native thinking flag wiring + VLM projector loading, media ingestion, and VLM streaming generation.
gguf_lib/src/main/cpp/CMakeLists.txt	Makes LLAMA_DIR configurable, conditionally configures KleidiAI, and builds/links VLM engine sources.
gguf_lib/build.gradle.kts	Expands ABI filters to include x86_64.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 25c50c10d3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[Siddhesh2377]

Hey man closing this PR, as the STT and TTS support is added via onnx not gguf, as gguf is very memory heavy