Enable Connection Metadata Navigation in Dataflow MCP Tools

Title: Enable Connection Metadata Navigation in Dataflow MCP Tools

Problem:
Currently, exploring a dataflow's connections requires:

1. Calling get_decoded_dataflow_definition to extract raw connection JSON
2. Parsing embedded IDs (ClusterId, DatasourceId, lakehouseId, workspaceId) from nested strings
3. Writing custom M queries to enumerate tables/schemas within those connections

There's no direct way to answer "what data sources does this dataflow connect to, and what's inside them?"

Proposed Solution:
Add a get_dataflow_connections tool that returns:

• Connection type (Lakehouse, SQL, Web, etc.)
• Resolved metadata (workspace name, lakehouse/database name, not just GUIDs)
• Available tables/entities with column schemas
• Connection credentials status (configured/missing)

Example Output:

{
  "connections": [{
    "type": "Lakehouse",
    "workspaceName": "DIExplorations",
    "lakehouseName": "olist_lakehouse",
    "tables": [
      {"name": "customers", "columns": [...], "rowCount": 99441},
      {"name": "orders", "columns": [...], "rowCount": 99441}
    ]
  }]
}

Value:

• Eliminates manual M query exploration
• Enables agents to reason about data lineage without trial-and-error
• Foundation for schema-aware query generation

Start by tracing the existing get_decoded_dataflow_definition tool and the surrounding Dataflow MCP tool entry points. Define how get_dataflow_connections should resolve connection metadata, enumerate schemas, and report credential status; done means it returns the specified connection details for supported data sources with tests covering the output.