Add blocking deployments

[ludfjig]

We should add something like a -wait or -detach:false flag so that we don't execute the next job until the current is finished. This will require monitoring/polling the evalId and depoymentIds with the api.