e2e CI tests

The repo should do an end2end test run based on the actual action's Terraform/Ansible code.

We can run a full `provision->deploy->destroy` pipeline on every PR (it's 10mins). As far as I remember, external collaborators won't be able to run the CI on our behalf so we're good as long as BitOps doesn't leak (#36).

If that feels risky, we can use `/commands` comments in GH PRs for more granular control the CI runs for testing to avoid overusing the compute resources. Alternatively, use the [[ci skip]](https://docs.github.com/en/actions/managing-workflow-runs/skipping-workflow-runs) flags.

Are we confident in our work, ready to dogfood own GH Action and test deploy ST2 instance based on defaults to EC2? This will help verifying the action runs end-to-end on every new feature/change, avoid regressions and also allows testing the changes directly in this repo, rather than using another middle repo, speeding up the development and feedback loop.

まず、リポジトリに既存する GitHub Actions workflows と、StackStorm インスタンスのプロビジョニング、デプロイ、破棄に使用される Terraform/Ansible パスを特定します。すべての PR で実行をトリガーするのか、制御されたコマンド／スキップ機構を使うのかを定義します。完了の定義は、デフォルトの EC2 デプロイが BitOps を漏えいさせたりコンピュートリソースを浪費したりすることなく、エンドツーエンドで完了することです。

索引モデルが issue の本文から書いたものです。