Slurmctld priority of zero treated like job hold (not scheduled).
Preserve priority order even if incomming job could use idle resources.

 o slurmd: if there is room in cmdline, change argv to show
   what job a slurmd is managing.

interrupts were resetting timeout in I/O functions with timeout.

   before opening output files. (Thus, seteuid_and_chdir() renamed
   to _drop_privileges()/_reclaim_privileges())
 o io.c : add srun info objects to tasks' srun list as they attach,
   so new sruns get task exit messages.

 o reattach.c : fixed search for matching job id.

   support documented uses of --input,output,error)
 o Allow premature termination of srun (i.e. when IO is not yet
   complete) by insistent user.
 o added xstrfmtcat() and xmemcat() to xstring.c

Perform allocate and step create independently.

call REQUEST_IMMEDIATE_RESOURCE_ALLOCATION. This immediate flag also
applies to the slurm_allocate_and_run_job_step() function.

between functions.

Move batch_flag in slurmctld internal database from job details to primary
   job record table.
Save/restore kill_job_on_step_done flag (for steps initiated with allocate
   RPC).

If set, the job step termination is treated like job termination (i.e. the
nodes are released and the job state set to complete).

Change pack32_array use 32 bit counter (rather than 16 bits).
Clean-up of timeout handling in srun.

Change pack32_array use 32 bit counter (rather than 16 bits).

   still remain.
 o cbuf.[ch], list.[ch] : Checked in new revisions from lsd-tools
 o srun/io.c,msg.c : added io_thr_create(), msg_thr_create() to simplify
   code in main().
 o Other minor fixes/changes

the user's job.

with step id)

allocation unless explicitly specified by user. It was not using an
appropriate task distribution for each job step, as it should based
upon node and task counts.

initiation.
Establish limit of MAX_TASKS_PER_NODE and add matching error code.