[ECS] [Service Discovery]: Support Singleton Deployments

**Tell us about your request**
Some applications require single-instance deployments. This seems to be in conflict with Service Discovery however.

**Which service(s) is this request for?**
ECS/Fargate

**Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard?**
Some applications expect to be the only one in existence by (for example) maintaining a file lock. (example: Sonatype Nexus). For this we can specify a `AWS::ECS::Service` with a `DeploymentConfiguration` that has `MaximumPercent: 100` and `MinimumHealthyPercent: 0`. This shuts down the existing container, then launches a new one.

However some peer applications may want direct access to this service so we install a `AWS::ServiceDiscovery::Service` on top. Upon testing this, I noticed ECS had launched a completely new service without having shut down the first. The log of the new one showed it discovering an existing instance and thus terminating.

So I rolled back, then commented out the ServiceDiscovery bits again and re-deployed. This worked again in the expected manner of shut-down then launch-new. So I conclude the two are mutually exclusive, at least for now.

**Are you currently working around this issue?**
We have to use a third container (traefik) to accept the traffic and route to the singleton back-end. The sibling application can then use the service DNS of traefik to target it's request.

Start by reviewing the ECS/Fargate deployment configuration using MaximumPercent: 100 and MinimumHealthyPercent: 0, then compare its behavior with the AWS::ServiceDiscovery::Service setup described here. Done means singleton deployments can use service discovery without launching overlapping instances or requiring the Traefik workaround.

Written by the indexing model from the issue text.