docs: remove AI-writing tells from model card (style-only; frontmatter untouched)

docs: WorldPolicies catalog badge

docs: badge row instead of blockquote for the eval path (renders cleaner on the Hub)

docs: surface Inspect Robots / WorldEvals eval path at top of card

docs: comply with worldevals model-card standard (license: gemma via PaliGemma backbone, metadata, safety, losses, provenance)

Upload README.md with huggingface_hub