Improve dashboard walkthrough documentation

[BlackthornEmpire]

Goal

Improve the dashboard walkthrough documentation so first-time users can quickly understand what each dashboard panel means and how to interpret retrieval risk states.

Why this matters

The README already includes dashboard screenshots, but users still need clear written guidance explaining what they are looking at.

This issue should improve the documentation around the existing dashboard views rather than adding unnecessary new UI work.

Suggested files

• docs/dashboard_walkthrough.md
• README links only if needed

Useful scope

Improve the written walkthrough for:

• compact view
• expanded view
• pipeline stages
• red fault states
• retrieval quality gauge
• candidate evidence panel
• bottleneck reporting
• conflict and density risk states
• local adviser mode
• local event logging

Expected outcome

A first-time user should be able to open the dashboard walkthrough and understand:

• what each panel does
• what healthy vs risky retrieval looks like
• what red fault states mean
• how candidate evidence should be inspected
• how bottleneck suggestions should be interpreted
• when retrieval should be blocked before generation

Acceptance criteria

• Improves explanation of existing dashboard views
• Does not duplicate screenshots already shown in the README unless needed
• Explains compact and expanded modes clearly
• Explains red fault states and retrieval risk indicators
• Explains candidate evidence inspection
• Explains bottleneck reporting and fix suggestions
• Explains local adviser mode
• Keeps the documentation practical and concise

Notes

This is a documentation improvement task, not a dashboard redesign.