[Feat]: sync llama.cpp/llama.rn

[a-ghorbani]

Description
Sync llama.cpp and llama.rn

• adds dry sampler
• adds static method for read model info from gguf (very useful for local models)
• fixes issue with simulators (Bug: SwiftUI example does not work on simulator. ggml-org/llama.cpp#10089)
  • should resolve [Bug]: Can't load models (iOS simulator) #94