Provide a supported standalone dashboard mode against an existing engine

[Regis-RCR]

Summary

I would like a persistent web dashboard that stays up on its own, independent of memtrace start and of any editor MCP session. Today neither command fits that cleanly:

• memtrace start owns the whole workspace: it spawns or adopts the engine, indexes, arms watchers, and can reclaim ownership. Serving the dashboard is a small part of that.
• memtrace mcp --ui does attach the dashboard to an existing engine, which is close to what I want, but it is an MCP server built for an editor or agent session, so running it as a long-lived UI daemon seems more like a workaround than a supported mode.

What I would like

A documented, supported way to serve the dashboard only, attached to an existing engine endpoint, without indexing, watching, or taking ownership. For example:

• a dedicated command such as memtrace ui --engine http://127.0.0.1:50051, or
• documented support for memtrace mcp --ui (or a --ui-only variant) as a standalone dashboard service.

I would also like that standalone mode to honor a configurable bind address such as MEMTRACE_UI_HOST, as raised in #63.

Why

• A persistent status dashboard that survives restarts, without leaving an editor session open to keep it alive.
• A clean split between the dashboard and the indexing and ownership behavior of start, so the engine and dashboard can be managed and restarted independently.

This ties back to #63, since mcp --ui is currently the only path that attaches the dashboard to an already-running engine.

[blocksorg]

Mention Blocks like a regular teammate with your question or request:

@blocks can you add technical details to this issue?
@blocks make the following changes ...
@blocks create an issue from what was mentioned in the following comment ...
@blocks explain the following code ...
@blocks are there any security or performance concerns?

Run @blocks /help for more information.

Workspace settings | Disable this message