Refactor: Implement Robust Startup Sequence and Health Checks in Memory Core

The Memory Core MCP server entry point (mcp-stdio.mjs) has been refactored to implement a more robust and informative startup sequence, leveraging the recent enhancements to the HealthService.

Key Changes:

1. Proactive Startup Health Check: The server now performs a comprehensive health check immediately upon starting.
2. Detailed Status Logging: Based on the health status (healthy, degraded, unhealthy), the server logs detailed, user-friendly messages, including actionable tips for resolving issues (e.g., how to start the ChromaDB if it's not running).
3. Automatic Session Summarization: If the health check passes and the GEMINI_API_KEY is configured, the server automatically triggers a background task to summarize any previously unsummarized sessions.
4. Tool Execution Gatekeeper: The main tool handler now calls HealthService.ensureHealthy() before executing most tools. This prevents operations from failing with cryptic errors if a dependency is down.
5. Precise Health Check Exemptions: The logic for skipping the health check has been refined from a broad string match (.includes('database')) to a specific allow-list (['healthcheck', 'start_database', 'stop_database']), making it safer and more explicit.

These changes improve the server's resilience, provide better feedback to the user, and automate routine maintenance tasks.