[TUI] Make critical session lists easier to scan and act on

[luoyuctl]

Problem

The critical session list contains useful data, but it is hard to scan quickly. The table, driver summary, and selected session all use similar visual weight.

Current screenshot:

Expected direction

• Make source/model/anomaly drivers visually distinct from the table.
• Highlight the selected session's reason: cost, duration, failure, or hanging.
• Add stronger visual treatment for cost, tokens, duration, and issue type.
• Reduce line noise and keep columns aligned for smaller terminal widths.

Validation notes

Captured from a real local run, not demo data.

[luoyuctl]

Event: Handoff
Actor: Product
Scope: #144
State change: none; labels now route this as lane/product,status/ready-for-agent
Evidence: Critical-list screenshot and issue body show drivers, selected reason, and table rows compete visually.
Next owner: Product

Handoff to: Product
Reason: This is TUI triage scanability and action clarity.
Expected output: Critical session list with clearer driver hierarchy, selected-session reason, and aligned cost/token/duration/issue signals.
Acceptance check: Before/after capture at normal and narrower terminal widths; go test ./...; no report output contract changes.

[luoyuctl]

Parser & Implementation Agent is taking this.

[luoyuctl]

Implementation PR: #158

Validation run:

• go test ./internal/tui -run 'TestList(DriverSummaryShowsDominantGroupsAndCounts|ViewShowsSelectedSessionSummary|ViewSelectedSummaryFitsNarrowCriticalList)|TestViewsRenderWithinTerminalWidth|TestViewsFitVerySmallTerminalBounds' -count=1\n- go test ./...\n- go build -o /tmp/agenttrace-parser-check ./cmd/agenttrace\n- /tmp/agenttrace-parser-check --doctor\n- /tmp/agenttrace-parser-check --demo --overview -f json\n\nNotes:\n- Normal-width capture covered at 120 columns.\n- Narrow critical-list capture covered at 80 columns with max-width assertion.\n- No report output contract changes.