[Feature Request] Add per-MCP-server restart command and health monitoring

Bug Description
MCP servers get stuck frequently with no way to restart them individually.

Problem:
The Linear MCP server occasionally becomes unresponsive (hangs on tool calls).
The only current solution is to restart the entire Claude Code session, losing
conversation context.

Expected behavior:
A command like /mcp restart <server-name> to restart a specific MCP server
without closing the session.

Workaround attempted:

• Killing the MCP process manually (unclear if Claude Code auto-reconnects)
• Using /mcp menu (no restart option available)

Impact:
High friction when using MCPs for extended sessions. Users lose context and
have to re-establish conversation state.

Suggestion:

1. Add /mcp restart <name> command
2. Add health checks with auto-reconnect on timeout
3. Show MCP server status (connected/hung/disconnected) in /mcp

Environment Info

• Platform: darwin
• Terminal: ghostty
• Version: 2.1.5
• Feedback ID: d9724841-e234-48dd-b0ea-f7b54719370d

Errors

[]