Remove council_query tool to force stage-by-stage usage

The all-in-one council_query tool returned everything as a single blob,
preventing the iOS app from displaying individual model responses as
separate content blocks. By removing it, the agent must now use:

- council_stage1_collect (individual responses)
- council_stage2_rank (peer rankings)
- council_stage3_synthesize (final synthesis)

Each stage call generates a separate tool_result, enabling granular
display in the UI.

mcp_server/server.py

@@ -59,55 +59,7 @@ async def api_request(
 
 
 # ============================================================================
-# HIGH-LEVEL COUNCIL TOOL
-# ============================================================================
-
-@mcp.tool()
-async def council_query(
-    query: str,
-    conversation_id: str | None = None
-) -> dict[str, Any]:
-    """
-    Run the complete LLM Council 3-stage deliberation process.
-
-    Stage 1: Query all council models in parallel for individual responses.
-    Stage 2: Each model anonymously ranks the other responses.
-    Stage 3: Chairman model synthesizes a final answer based on all inputs.
-
-    This process typically takes 1-3 minutes depending on model response times.
-
-    Args:
-        query: The user's question to deliberate on
-        conversation_id: Optional existing conversation ID for multi-turn context
-
-    Returns:
-        Complete council response with all 3 stages and metadata including:
-        - conversation_id: The conversation ID used
-        - stage1: Individual model responses
-        - stage2: Peer rankings with aggregate scores
-        - stage3: Chairman's synthesized final answer
-        - metadata: Label mappings and aggregate rankings
-    """
-    # Create conversation if not provided
-    if not conversation_id:
-        conv = await api_request("POST", "/api/conversations", {})
-        conversation_id = conv["id"]
-
-    # Send message and get full council response
-    result = await api_request(
-        "POST",
-        f"/api/conversations/{conversation_id}/message",
-        {"content": query}
-    )
-
-    return {
-        "conversation_id": conversation_id,
-        **result
-    }
-
-
-# ============================================================================
-# INDIVIDUAL STAGE TOOLS
+# INDIVIDUAL STAGE TOOLS (council_query removed to force stage-by-stage usage)
 # ============================================================================
 
 @mcp.tool()