[NPU] Expose parameter to control blob / IR save logic

[rnwang04]

Description

1. Why the change?

https://github.com/analytics-zoo/nano/issues/1833#issuecomment-2635643123
https://github.com/intel-analytics/llm.cpp/pull/807

2. User API changes

Expose two parameter keep_ir and compile_blob for AutoModelForCausalLM.from_pretrained, keep_ir is default to False and compile_blob is default to True.
In C++ convert.py script,

• default blob convert: python convert.py --repo-id-or-model-path D:\llm-models\Llama-3.2-3B-Instruct --save-directory D:\Llama-3.2-3B-Instruct-npu-q4_0-blob
• ir convert: python convert.py --repo-id-or-model-path D:\llm-models\Llama-3.2-3B-Instruct --save-directory D:\Llama-3.2-3B-Instruct-npu-q4_0-ir --keep-ir --disable-compile-blob

3. Summary of the change

• Expose two parameter keep_ir and compile_blob for AutoModelForCausalLM.from_pretrained, keep_ir is default to False and compile_blob is default to True
• update C++ convert.py script
• Remove unnecessary bin files to save disk space

4. How to test?

• Unit test: Please manually trigger the PR Validation here by inputting the PR number (e.g., 1234). And paste your action link here once it has been successfully finished.

[plusbang]

LGTM