Update Dataflow

Issue: Add update_dataflow_definition tool

Labels: enhancement, dataflow-management, priority:high, automation-blocker

Problem: Dataflow destinations (Lakehouse tables, warehouses) cannot be configured programmatically. Users must manually configure write locations in Fabric UI, breaking automation workflows. Current tools can create dataflows and read definitions but cannot modify them.

Proposed Solution: Implement update_dataflow_definition tool that:

• Accepts workspaceId, dataflowId, and definition payload (structured JSON or base64-encoded)
• Calls Fabric REST API: PATCH /v1/workspaces/{workspaceId}/dataflows/{dataflowId}/updateDefinition
• Supports updating queryMetadata.json (destinations), mashup.pq (M code), and connection references
• Optional helper: set_dataflow_destination wrapper that modifies only destination config without full definition manipulation

Acceptance Criteria:

• Tool successfully updates dataflow definition via Fabric API
• Supports setting Lakehouse table as destination (most common use case)
• Handles base64 encoding of mashup sections correctly
• Returns validation errors if definition is malformed
• Documentation includes end-to-end example: create dataflow → add connection → set destination → execute
• Does not overwrite displayName/description unless explicitly provided
• Works with definitions retrieved from get_decoded_dataflow_definition

Start by reading the existing get_decoded_dataflow_definition flow and the Fabric PATCH endpoint for /v1/workspaces/{workspaceId}/dataflows/{dataflowId}/updateDefinition. Trace how definitions, base64-encoded mashup sections, and connection references are represented. Done means updates validate, preserve unspecified metadata, support Lakehouse destinations, and cover the documented end-to-end workflow.