OpenVINO Backenb: remove changes in llama.cpp

[zhaixuejun1993]

This pull request removes the org_src member from the ggml_tensor struct and updates all related logic throughout the codebase. This change simplifies the tensor structure and removes checks and assignments related to org_src in both the OpenVINO decoder and backend scheduling code.

Tensor structure simplification:

• Removed the org_src pointer from the ggml_tensor struct in ggml.h, reducing the size of the padding array accordingly.
• Updated the tensor initialization in ggml.c to remove the initialization of org_src.

Code cleanup and logic updates:

• Removed all assignments to and checks for org_src in backend scheduling logic in ggml-backend.cpp.
• Updated the OpenVINO decoder (ggml-decoder.cpp and ggml-decoder.h) to remove logic that referenced or depended on org_src, simplifying conditions and removing unused code. [1] [2] [3]

These changes reduce complexity and potential confusion around tensor source tracking, making the codebase easier to maintain.## Overview

Additional information

Requirements

• I have read and agree with the contributing guidelines
• AI usage disclosure: