Improve logging when withReuse() is specified and a container cannot be reused.

I would like to see a warning logged everytime a container is started withReuse(true) and the hash does not match.

Ideally the warning should contain details of the existing running containers, their container ids and mismatched hashes.

And if possible it would be REALLY helpful to have some details about why the hash didn't match. I realize from reading the code that this isn't possible currently given that the hash is computed from the CreateContainerCmd but maybe some comparisions would be performed of the InspectContainerResponse post container start, as compared to the existing containers.. It would probably make sense to log this level of detail at DEBUG or TRACE levels.

Right now we are facing issues where it's very easy to trigger new containers to be created when the intent was to reuse an existing container, and it's very easy to miss this happening in the logs unless you are paying special attention to it.

And once you've spotted it, it's very hard to troubleshoot why a new container was created vs an existing one being reused.

Sometimes it's some init-data that was updated so the files hash has changed, othertimes, it's something more subtle like a host bind path having a trailing / but otherwise being exactly the same.. This last issue has other implications, which probably warrants it's own ticket, but at the moment it's hard to gather all the evidence easily.. hence this ticket.

まず、withReuse(true) をコンテナ再利用の経路に沿って追跡し、次に CreateContainerCmd のハッシュが既存の実行中コンテナとどのように比較されるかを調べます。利用可能な InspectContainerResponse の詳細とロギング箇所を確認します。再利用の不一致によってコンテナ ID とハッシュを含む警告が出力され、利用可能な場合は DEBUG または TRACE でより詳細な比較情報が出力されれば完了です。