fix: implement ping method in MCP server

[arnoldwender]

What does this PR do?

Adds a ping handler to the MCP stdio dispatch loop.

MCP clients such as AnythingLLM v1.12.0 send a ping health-check immediately after connecting. Without a handler the server returns {"code": -32601, "message": "Unknown method: ping"}, which causes the client to kill and restart the server in an infinite loop.

The fix adds a single branch that responds with an empty result object, as specified by the MCP protocol:

{"jsonrpc": "2.0", "id": <req_id>, "result": {}}

Fixes #584

How to test

1. Connect AnythingLLM v1.12.0 (or any MCP client that sends ping) to the server
2. Confirm the server stays up instead of restarting in a loop

Or directly:

echo '{"jsonrpc":"2.0","id":1,"method":"ping","params":{}}' | python -m mempalace mcp run
# Expected: {"jsonrpc": "2.0", "id": 1, "result": {}}

Checklist

• Tests pass (python -m pytest tests/ -v)
• No hardcoded paths
• Linter passes (ruff check .)

[arnoldwender]

Closing — PR #600 by @grtninja was already merged with the same fix. I wasn't aware it had landed. Sorry for the noise.