Add SciForge visualization pipeline

[PatrickPeiyuHe]

Summary

Closes #6 by adding a full SciForge visualization pipeline and wiring it into the completed validation loop.

• Adds agents/visualization_agent.py as the dedicated automated visualization module.
• Generates programmatic experiment figures from experiment_iterations and experiment_runs:
  • metric trajectory
  • baseline vs proposed comparison
• Generates literature/result charts from the results table when contextual data exists.
• Generates conceptual diagrams-as-code from deep_insights content:
  • approach / motivation overview
  • method architecture diagram
  • knowledge graph subgraph from graph_relations
• Writes figures under the experiment workspace figures/ directory.
• Emits SVG/PDF/render-source/DOT artifacts plus figure_manifest.json.
• Registers renderable figures in experiment_artifacts with artifact_type='plot' and source/manifest files as source_data.
• References generated figures in final_report.md when present, or writes a figure_references.md sidecar.
• Keeps visualization generation and artifact registration non-blocking.

[cla-assistant]

All committers have signed the CLA.

[PatrickPeiyuHe]

I have read the CLA.md and agree to the terms.

Could a maintainer please add @PatrickPeiyuHe to cla-signers.json so the repository CLA workflow can pass?