Terraform apply fails first, then works 2nd time in some cases

Sometimes a `terraform apply` run does some things successfully but fails before completing. Then I rerun it again immediately without any other changes/actions and it succeeds. My strong hunch about this is that most likely my resource dependencies are not modeled correctly in terraform. I have seen this in at least 2 specific cases.

1. Fresh install into a blank slate AWS account.

```
aws_api_gateway_deployment.hexagonal-lambda-dev: Error creating API Gateway Deployment: BadRequestException: The REST API doesn't contain any methods
status code: 400, request id: redacted
```

2. Creating a new lambda function.

```
Error applying plan:

1 error(s) occurred:

* aws_iam_role_policy_attachment.post-up: 1 error(s) occurred:

* aws_iam_role_policy_attachment.post-up: [WARN] Error attaching policy arn:aws:iam::REDACTED:policy/terraform-redacted to IAM Role post-up: NoSuchEntity: Role not found for post-up
status code: 404, request id: redacted
```

Inspecione a configuração do Terraform para aws_api_gateway_deployment, os métodos de API, a função Lambda e aws_iam_role_policy_attachment.post-up. Primeiro, reproduza um apply inicial em uma conta AWS vazia e compare o grafo de dependências dos recursos com a segunda execução; considera-se concluído quando o primeiro apply termina sem nenhum dos dois erros relatados.

Escrita pelo modelo de indexação a partir do texto da issue.