Parameterize trust_remote_code for quark_visitai

[shaahji]

Parameterize trust_remote_code for quark_visitai

trust_remote_code was being defaulted to True raising security concerns. Replacing the default with user driven value thru' the config.

Checklist before requesting a review

• Add unit tests for this change.
• Make sure all tests can pass.
• Update documents if necessary.
• Lint and apply fixes to your code by running lintrunner -a
• Is this a user-facing change? If yes, give a description of this change to be included in the release notes.

(Optional) Issue link

[Copilot]

Pull request overview

This PR makes Quark Vitis AI quantization safer by removing hard-coded trust_remote_code=True in Hugging Face model/tokenizer loading and instead deriving the value from user configuration (pass config and/or the model handler load kwargs).

Changes:

• Add trust_remote_code parameters to Quark Vitis AI LLM model/tokenizer preparation helpers and propagate into from_pretrained(...).
• Introduce a trust_remote_code pass config knob in QuarkQuantizationVitisAI and plumb the resolved value into the torch quantization runner.
• Update the quantization runner to pass trust_remote_code through to model/tokenizer loading.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
olive/passes/quark_vitisai/torch/language_modeling/llm_utils/model_preparation.py	Parameterizes trust_remote_code for tokenizer/model loads (and tokenizer re-load in save_model) instead of forcing True.
olive/passes/quark_vitisai/torch/language_modeling/llm_ptq/quantize_quark.py	Threads trust_remote_code into get_model/get_tokenizer calls used by Quark torch quantization.
olive/passes/quark_vitisai/quark_quantization_vitisai.py	Adds pass-level trust_remote_code config and resolves it from model load kwargs with an explicit override path.