Refactor job resources/constraints
We currently have the following resource spec:
- <type>://<host>[:<port>]/<schedd>/<lrms-system-name>/<queue-name> for condor backend (condor-submit)
- <host>:<port>/cream-<lrms>-<queue> for cream backend (glite-ce-job-submit)
- <host>:<port/2811>/nordugrid-<lrms>-<queue> for arc backend (arcsub)
In addition, there are number of other job requirements apart from queue (like walltime, cputime, memory, runtimeenv, etc.) that all go in the JDL and should be configurable via some common spec across all backends.
One option would be to use common resource spec matching the one for condor without queue and then have additional set of arguments like --jdl-=value for the other constraints.