Wrong desired count

### Summary
Desired count is set wrong when an appautoscaling policy is triggered during deployment

### Description
Desired count is set to the sum of `ACTIVE + PRIMARY + SCALING ADJUSTMENT` when a policy is triggered while a deployment is running.

### Expected Behavior
Set correct desired count

### Environment Details
ecs instance ami amzn2-ami-ecs-hvm-2.0.20190614-x86_64-ebs (ami-04a084a6d17d9816e)

agent version 1.29.0

##### Deployment config
Minimum healthy percent: 100
Maximum percent: 200

##### AutoScaling config
Minimum tasks: 6
Maximum tasks: 24

SCALE-UP-service/prod/service-prod: CPUUtilization >= 99
Policy type: Step scaling
For alarm: service-prod_CPUUtilization
Take the action:
Add 1 tasks when 99 <= CPUUtilization

SCALE-DOWN-service/prod/service-prod: CPUUtilization >= 99
Policy type: Step scaling
For alarm: service-prod_CPUUtilization
Take the action:
Remove 1 tasks when <= CPUUtilization 99

### Supporting Log Snippets
(from ECS/My Service/Events)
autoscaling activity before deployment
```
2019-07-10 15:29:44 +0200 Message: Successfully set desired count to 10. Change successfully fulfilled by ecs. Cause: monitor alarm ECS_Service-service-prod_CPUUtilization in state ALARM triggered policy SCALE-UP-service/prod/service-prod
2019-07-10 15:29:48 +0200 service service-prod has started 1 tasks: task 269a6d27375c4d8b8def40061a341456.
2019-07-10 15:30:02 +0200 service service-prod has reached a steady state.
```
deployment begins here
```
2019-07-10 15:31:03 +0200 service service-prod has started 10 tasks
```
autoscaling policy is triggered
```
2019-07-10 15:31:03 +0200 Message: Successfully set desired count to 21. Change successfully fulfilled by ecs. Cause: monitor alarm ECS_Service-service-prod_CPUUtilization in state ALARM triggered policy SCALE-UP-service/prod/service-prod
2019-07-10 15:31:50 +0200 service service-prod has started 3 tasks: task f0cc7d528c294519984069c278fd0582 task 80e267e653f84f9e836540e17f394472 task e381f9755bc8406d8d8a3f9cce61e9b2.
2019-07-10 15:32:04 +0200 service service-prod was unable to place a task because no container instance met all of its requirements. The closest matching container-instance da22b00b-0b94-4c9f-a794-74859595e552 has insufficient CPU units available. For more information, see the Troubleshooting section.

... more of started/stopped tasks till deployment is completed
```

No repository files or tests are identified. Start by reviewing the reported ECS deployment and autoscaling event sequence, including the desired-count changes from 10 to 21, then determine the expected count during deployment; done means the service sets the correct desired count without adding the deployment adjustment twice.

Written by the indexing model from the issue text.