[Feature Request] support queue mode MCP server configuration

[Marshall-Qiu]

Summary

Add support for configuring dedicated MCP (Model Context Protocol) webhook deployments and ingress routing to properly handle MCP Server Trigger nodes in queue mode with multiple webhook replicas.

Documentation

https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-langchain.mcptrigger/#limitations

Problem Statement

• The MCP Server Trigger node relies on Server-Sent Events (SSE) and streamable HTTP connections, which require the same server instance to handle persistent connections. This creates issues when running n8n in queue mode with multiple webhook replicas, as documented in the n8n documentation:
• Single webhook replica: MCP Server Trigger node works as expected
  Multiple webhook replicas: All /mcp* requests must be routed to a single, dedicated webhook replica

Currently, the n8n Helm chart doesn't provide built-in support for this configuration, requiring manual setup of:

1. A separate webhook deployment dedicated to MCP requests
2. Custom ingress routing for /mcp/ and /mcp-test/ paths
3. Service configuration to target the dedicated MCP webhook instance

Proposed Solution

Enhance the Helm chart to support dedicated MCP webhook deployments through new configuration options:

1. New Values Structure
   Add a new mcpWebhook section to values.yaml
2. Ingress Enhancement
   Extend the existing ingress template to automatically include MCP routing when mcpWebhook.enabled: true