Currently the image.run command runs in detached mode (-d). Give that the first parameter is args to docker and it is required, users should pass this as -d if they want. Currently you have to pass empty string if you want to use docker run.
[line 142 here](https://github.com/jenkinsci/docker-workflow-plugin/blob/master/src/main/resources/org/jenkinsci/plugins/docker/workflow/Docker.groovy)
Example:

Today it would return a detached container:

node {
   docker.image("ubuntu").run('', 'npm test')
}

How it could be:

node {
    docker.image("ubuntu").run('-d', 'npm test')
}

Using attach for example:

node {
    docker.image("ubuntu").run('-a STDOUT', 'npm test')
}

This second example today is not even possible, since -a STDOUT -d result in a conflict and docker ends in exception.