Add comprehensive Databricks deployment guide

Documentation:
- Explains MCP server vs Databricks notebook usage
- 4 deployment patterns with detailed examples
- Local MCP server for Claude Desktop integration
- Databricks notebooks for native workflows
- Databricks Jobs for production automation
- Future Databricks Apps considerations

Key Concepts:
- MCP server uses stdio - not suitable for notebooks
- Databricks should use SDK directly (no MCP layer)
- Provides reusable Python functions for notebooks
- Comparison table for different deployment patterns

Code Examples:
- Provision share in notebook (complete workflow)
- Validate share readiness function
- Scheduled data product deployment
- Databricks Job configuration JSON
- Reusable module structure

Recommendation:
- Local machine: Use MCP server with LocalDatabricksClient
- Databricks notebooks: Use direct SDK calls
- Production: Use Databricks Jobs with scheduled notebooks
- NOT recommended: Running MCP server in Databricks

This addresses the architectural question of where and how
to deploy the MCP server vs using the underlying functionality
directly in Databricks environments.

Author: MarioDeFelipe