(aws-ecs): Deleting a stack with a Cluster and an ASG capacity provider fails

### What is the problem?

If a stack contains a ECS cluster with an ASG capacity provider, deleting the stack will fail due to an apparent circular dependency.

### Reproduction Steps

Minimal stack:

```
class ClusterWithCapacity(cdk.Stack):
def __init__(
self,
scope: Construct,
id: str,
**kwargs,
) -> None:
super().__init__(scope, id, **kwargs)
vpc = ec2.Vpc(self, "vpc")
ecs_cluster = ecs.Cluster(self, "cluster", vpc=vpc)

asg = autoscaling.AutoScalingGroup(
self,
"asg",
vpc=vpc,
instance_type=ec2.InstanceType.of("t3-small"),
machine_image=ecs.EcsOptimizedImage.amazon_linux2(),
)

capacity_provider = ecs.AsgCapacityProvider(
self,
"capacity_provider",
auto_scaling_group=asg,
)

ecs_cluster.add_asg_capacity_provider(provider=capacity_provider)
```

Deploy the stack and try deleting it.

### What did you expect to happen?

The stack is deleted.

### What actually happened?

The capacity provider fails to delete with the following error:
`Resource handler returned message: "Invalid request provided: Cannot delete a capacity provider while it is associated with a cluster."`

### CDK CLI Version

2.15.0

### Framework Version

_No response_

### Node.js Version

17.6.0

### OS

MacOS

### Language

Python

### Language Version

_No response_

### Other information

This hints that the cluster needs to be deleted first, but the capacity provider actually depends on the cluster.

Creating an explicit dependency `ecs_cluster.node.add_dependency(capacity_provider)` results in a circular dependency.

Start by reproducing the minimal Python stack using the ECS Cluster, AutoScalingGroup, and AsgCapacityProvider entry points shown in the issue, then inspect the corresponding aws-cdk ECS construct implementation. Done means the stack can be deleted without the capacity provider error or a circular dependency.

Written by the indexing model from the issue text.