Schema + viewer: bind concepts to nodes (per-node concept chips)

[EYH0602]

Background

Gap analysis vs the deployed ara-hub viewer (https://www.agenticresearch.sh/ara/AmberLJC/ara-paperbench/artifacts/speedrun/nanogpt-speedrun). This is a schema-widening task (parser + manifest + viewer), not viewer-only.

Problem

The hub binds glossary concepts to individual nodes — each node lists the concepts it introduces or uses (e.g. N03 → "Muon Optimizer"), rendered as chips in the detail pane. Our manifest has a global concepts list (Glossary modal) but no node→concept binding, so concept context is disconnected from the steps that use it.

Proposal

• Widen the node schema with an optional concepts: list of concept ids.
• Render concept chips in the detail pane; clicking a chip opens the Glossary panel focused on that concept (or a hover popover with the definition, matching the hub's .gterm/#gpop behaviour — pick one and note the choice).

Acceptance

• Node→concept bindings parse and appear in the manifest
• Concept chips render in the detail pane when present
• Chips link to the concept definition (modal focus or popover)
• Parser + viewer unit tests