Align VLM caption model, fix V2 ingest() example, document run_pipeline; misc README/Helm fixesUpdating files per bugs 5966185, 5966211, and 5966281

[kheiss-uwzoo]

Align VLM caption model, fix V2 ingest() example, document run_pipeline; misc README/Helm fixes

Summary

This change brings documentation and a pipeline default in line with the deployed VLM stack (Nemotron Nano 12B v2 VL), fixes a broken V2 Python quickstart snippet, and adds accurate library-mode reference for run_pipeline (parameters, defaults, and return types). Follow-up edits correct the stable-release branch link, image-caption fallback in the custom summarization pipeline, and a few Helm README details.

Internal refs: NVBug 5966185, 5966211, 5966281 (per commit message).

What changed

Documentation (docs/docs/extraction/)

Area Change

• VLM / caption Default and example model_name updated from nvidia/llama-3.1-nemotron-nano-vl-8b-v1 to nvidia/nemotron-nano-12b-v2-vl in python-api-reference.md and nv-ingest-python-api.md. Quickstart Profile table: vlm row now points to Nemotron Nano 12B v2 VL and the matching model card URL in quickstart-guide.md.
• V2 API Minimal example uses .ingest(return_full_response=True) and prints results[0]['metadata']['total_pages'] so the sample matches a list of per-job responses and top-level metadata in v2-api-guide.md.
• Library mode quickstart-library-mode.md: new run_pipeline section—parameter table (optional defaults per pipeline_runners.py), return-type table (RayPipelineInterface vs RayPipelineSubprocessInterface vs float), pointer to source docstring.
• Config
• config/custom_summarization_pipeline.yaml: image caption model_name fallback updated to nvidia/nemotron-nano-12b-v2-vl when VLM_CAPTION_MODEL_NAME is unset (matches default_pipeline.yaml / compose defaults).
• README / Helm (local only; commit or squash before merge)
• README.md: stable release branch link target set to NVIDIA/nv-ingest 26.03.
• helm/README.md: sample kubectl get pods output refreshed; grammar fix (“its process”); MIG example link corrected to mig/nv-ingest-mig-values.yaml.
• Testing / verification

[sosahi]

@jdye64 can you check these changes please?