on normal completion.

Now I will really be able to make some progress in squeue. ;-)

Also fixed some problems in the pack methods.

REQUEST_JOB_INFO now uses the job_info_request_msg_t instead of last_update so that in the future I will be able to request information for just one job.
Fixed a lot of bugs in the REQUEST_JOB_STEP_INFO message stuff.

Remove call to scheduling loop on failed job submit RPC (redundant).

added %m to a lot of debug messages

Added support for combined allocate resources and run job step API.
Changed allocate_tst.c to perform allocate_and_run RPC on first request.

Added support for combined allocate resources and run job step API.

moved things around and cleaned up things

Begining of cleanup of slurmd