Expose vllm_init_kwargs in semantic dedupe tutorials

[sarahyurick]

No description provided.

[greptile-apps]

Greptile Summary

This PR updates the semantic deduplication tutorials to show vLLM initialization options.

• Adds GPU memory utilization kwargs to the end-to-end workflow example.
• Adds the same vLLM option to the step-by-step stage setup.
• Updates the secrets baseline line numbers for the changed notebooks.

Confidence Score: 5/5

This looks safe to merge.

• No blocking issues found in the changed code.

Important Files Changed

Filename	Overview
tutorials/text/deduplication/semantic/semantic_e2e.ipynb	Adds embedding_vllm_init_kwargs to the workflow example.
tutorials/text/deduplication/semantic/semantic_step_by_step.ipynb	Adds vllm_init_kwargs to the explicit embedding stage setup.
.github/workflows/config/.secrets.baseline	Updates notebook line numbers in the secrets baseline.

_{Reviews (2): Last reviewed commit: "update secret" | Re-trigger Greptile}