[POC] Speed-up hash calculation when ov::model used

[praasz]

Details:

• Use model attributes to bind model blob_id with model path and custom transformation pipeline by ID modifier
• The GenAI for e.g. LLM pipeline should set model attributes to:

 model.get_rt_info()["__model_path"] = std::filesystem::path("model_path");
 model.get_rt_info()["__model_id_modifier"] = "Custom modifier for transformation pipeline";

Tickets:

• Slow model loading time comparing to optimum OV model openvino.genai#2526