[FEATURE] Export to GGUF and LLama.cpp

[vkkhare]

Describe the feature request
Bring support for Llama.cpp inferencing and benchmarking.

Describe the solution you'd like

• modelling_llama_skip.py changes for exporting to GGUF
• Add and dispatch inference to llama.cpp with sparse transformers GGUF
• update run_benchmark.py to support llama.cpp