Check for available resources before starting instances in jobs

It would be nice if jobs that declare they need additional resources can check that there is enough resources in openstack before trying to spin up VMs. This feature should also allow retrying up to a timeout limit before failing the job.

Our use case is that we have many integration jobs that spin up full labs of VMs before running the job. The current openstack plugin fails the build if the instances could not be spun up. We would prefer that the job confirmed that there was enough resources to spin up the additional instances before trying, and if the resources were not available check again periodically until I can allocate the resources. It should then fail the build only once reaching some timeout value like 30 minutes or so.