CLI/mcpdo: connect-timeout error tells the user to change "Server Settings", which only the web UI has #2615

Problem

When the connect-time timeout fires, the CLI and mcpdo tell the user to change a setting that only exists in the web UI:

$ mcp-inspector --cli --server-url http://127.0.0.1:6999/mcp --connect-timeout 1500 --method tools/list
{"error":{"code":"unreachable","message":"Connection timed out after 1500 ms. To accommodate a slower server, you may increase the timeout value in Server Settings."}}

The CLI has no "Server Settings". Its knobs are --connect-timeout <ms>, which both the one-shot CLI and mcpdo connect accept, and the server entry's connectionTimeout in mcp.json.

Where

connectionTimeoutMessage() in core/mcp/inspectorClient.ts (added for #2320). Its doc comment says "All three clients surface this text as-is: the web client's 'Failed to connect' toast, the TUI status line, and the CLI error envelope". The web wording is right for the web client and wrong for the others.

Expected

Each client names its own remedy. For example, the CLI and mcpdo would say to raise --connect-timeout or set connectionTimeout on the server entry. The TUI should be checked as well, since it says it surfaces the same text. One shape for the fix is a client-supplied hint (an optional parameter or a per-client mapping of the unreachable envelope), so core keeps the message and the remedy varies by client.

Reproduce

Point either CLI at a listener that accepts TCP connections and never responds (node -e 'require("net").createServer(()=>{}).listen(6999)'), with a short --connect-timeout. Found while smoke-testing #2613. The behavior predates that PR.