[model validation]: verify granite-4.0-tiny-preview inference

[Days-gone]

Purpose

verify the inference of granite-4.0-tiny-preview on vllm-metax.

Test Plan

using the code under: docs/models/samples/granite_4_0_tiny-preview.py

Test Result

(Optional) Documentation Update

update the supported document accordingly.

[gemini-code-assist]

Code Review

This pull request introduces a sample script to validate inference for the granite-4.0-tiny-preview model and updates the documentation to reflect its support. The changes are generally good, but I have identified a critical security issue and a high-severity dependency issue in the new sample script. My review includes suggestions to add a security warning and to remove an unnecessary dependency, which will improve the script's security and maintainability.

[gemini-code-assist]

The torchvision library is imported but appears to be unused for its core functionality. This adds a heavy and unnecessary dependency to what should be a minimal sample script. If this is intended as a workaround for a warning from another library, this should be documented. Otherwise, it should be removed to keep the example clean.