feat(eval): online-insight TUI get/list + #2029 insights column (P4)

[jariy17]

P4 — online-insight read-only TUI + #2029 insights column

Stacked on #2070 (base branch feat/online-eval-byo-role, P3). Read-only TUI only — no deploy/mutation paths.

New screens (src/handlers/eval/online-insight/)

• screen.tsx — OnlineInsightScreen (RouterScreen path ["agentcore","eval","online-insight"]).
• list/screen.tsx — OnlineInsightListScreen; reuses OnlineEvalPicker, reads via core.eval.listOnlineInsights.
• get/screen.tsx — OnlineInsightGetScreen + OnlineInsightGetJsonScreen via core.eval.getOnlineInsight. Detail shows insight count + clustering frequencies instead of the evaluator count.

Router + routes

• online-insight/index.tsx: added .use(withTuiOnEmptyFlagsAndArgs), .default(renderTui), .supportedTuiCommands("get","list"), and export { OnlineInsightScreen }.
• Root.tsx: added the 5 online-insight routes mirroring online-eval (menu · list · get→list redirect · get/:configId · get/:configId/json) and the screen imports.

#2029 — insights indicator column

Issue #2029: "online-eval imperative TUI table should include an 'insights' field that indicates whether that online eval has insights enabled."

• Added a has-insights column to the online-eval list table (and the online-insight list, same picker). Value = config.insights?.length > 0, rendered yes / -. Sourced from OnlineEvaluationConfigSummary.insights on the list summary — no extra API call.

Picker: parameterized, not forked

OnlineEvalPicker gained optional resourceLabel, queryKey, and loadPage props (defaulting to the online-eval behavior) so the online-insight list reuses the same component and columns. Columns/toRow are shared (identical for both lists) with the insights column baked into the default onlineEvalColumns, so no fork and no dead per-list column overrides.

Tests

• New online-insight.screen.test.tsx mirroring the online-eval screen test: menu (read-only commands only), picker (insights column yes/-, listOnlineInsights args, bare-get redirect, selection → detail, empty state), detail (sampling / execution / insight count / clustering frequencies, getOnlineInsight args, JSON view, retry).
• Added a online-eval imperative TUI table should include "insights" field that indicates whether that online eval has insights enabled #2029 insights-column case to online-eval.screen.test.tsx.
• bun test src/handlers/eval/online-insight + online-eval: 64 pass. Broader bun test src/components src/handlers/eval: 319 pass. tsc --noEmit: clean.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.37%. Comparing base (e51a676) to head (4de06d4).

Additional details and impacted files

@@             Coverage Diff              @@
##           refactor    #2072      +/-   ##
============================================
+ Coverage     97.36%   97.37%   +0.01%     
============================================
  Files           424      427       +3     
  Lines         25571    25695     +124     
============================================
+ Hits          24897    25021     +124     
  Misses          674      674              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.