Introduce lifecycle configuration to allow proper resource recreation

[szilveszter]

Unfortunately the current setup still has the issue that prevents us from replacing the API-GW resources when the Swagger file changes.

Fortunately someone else has chimed in with their solution, and this seemed to work during my tests:
hashicorp/terraform#6613 (comment)

Edit: After talking to @ryandub, and doing some more testing, it seems that some lifecycle configuration solves the original issue, and it looks cleaner than the intermediate stage approach.

Also changed the output of the module to reflect custom domains.

(My test case was only about changing the description variable of the API Gateway module, which affects the Swagger file.)

[ryandub]

For further context - looking at @szilveszter 's error, it appeared that Terraform was "destroying" the deployment resource, then creating a new one. From the provider code it looks like that process disassociates the stage from the deployment, then deletes the deployment (before then creating a new one) - however, @szilveszter was getting a 400 on some occasions where the stage did not fully disassociate before the attempt to delete the deployment. Hence the error message "BadRequestException: Active stages pointing to this deployment must be moved or deleted". Adding create_before_destroy here should cause a new deployment to be created and the stage assigned to that new deployment. Then, the delete on the old deployment resource should not have an issue.