configure-proxy can trigger duplicate nginx-proxy-manager startup

[vsilent]

Description

Running stacker agent configure-proxy website --domain mobile.syncopia.io --port 3003 appears to leave the stack in a bad proxy state on the next deployment: two nginx-proxy-manager containers show up, one keeps restarting, and deployment pauses because port 80 is already allocated.

This should be an idempotent proxy configuration change, but it can lead to duplicate proxy startup attempts and broken deployments.

Steps to Reproduce

1. Deploy a stack that uses proxy.type: nginx-proxy-manager.
2. Run:

   ./stacker agent configure-proxy website --domain mobile.syncopia.io --port 3003

3. Trigger the next deployment/reconfigure cycle.
4. Observe duplicate nginx-proxy-manager containers and/or a paused deployment caused by port 80 is already allocated.

Expected Behavior

Proxy reconfiguration should update the existing nginx-proxy-manager setup in place and keep exactly one proxy container/service definition bound to ports 80/81/443.

Actual Behavior

A second nginx-proxy-manager instance can be started or generated, causing port collisions and leaving deployment paused.

Acceptance Criteria

• Add a failing regression test that reproduces duplicate proxy-manager startup after configure-proxy followed by deploy.
• Add a BDD scenario for reconfiguring an existing nginx-proxy-manager-backed domain without creating duplicate proxy containers.
• Ensure deploy/configure-proxy remains idempotent and only one proxy service/container is produced for the stack.
• Prevent port 80 allocation conflicts caused by duplicate proxy-manager instances.