Viewer: render exhibit captions (description) in the result block

[EYH0602]

Background

Gap analysis vs the deployed ara-hub viewer (https://www.agenticresearch.sh/ara/AmberLJC/ara-paperbench/artifacts/speedrun/nanogpt-speedrun).

Problem

Exhibit.description is parsed into the manifest (crates/ara-core/src/manifest.rs:232) but dropped by the detail pane: render_detail only shows id · kind chips plus the body (crates/ara-viewer/src/detail.rs:552-575). The caption is only reachable as a hover tooltip on the chip (and there it's source/file, not the caption).

On real artifacts every table/figure carries a real caption (e.g. "Table 1: NanoGPT Speedrun Progression (21 Records)") — we show table1_speedrun_progression · table instead. The hub viewer renders captions prominently.

Proposal

Render exhibit.description as the caption above/below each rendered exhibit body (figcaption-style, mono muted per existing token conventions). Keep the chip row as-is.

Acceptance

• Exhibits with a non-empty description show it as a visible caption in the result block
• Exhibits without description render unchanged
• Unit test in detail.rs covering both cases

[EYH0602]

Fixed in #65.