This fixes a bug introduced in commit 52fe3de1
in the event the fork() call fails in slurmstepd.

Insure that a job is completely launched before trying to suspend it.
Previous logic would start suspend logic early in the life of the
slurmstepd process, after it's listening socket was open but before
the tasks were launched. This defers the suspend logic until after
all prologs and setup completes and the tasks are launched. This is
important in the case of gang scheduling, in which newly launched
jobs can be immediately suspended.
bug 2494

commit 75317972.

commit ff2c5b88

ff2c5b88

QOS flag affects.

deal with time limit with TRES limits.

to use this function with associations.

Bug 1976

If power save mode was configured along with PrologSlurmctld, then
when PrologSlurmctld completed, it was clearing the node's PowerUp
state flag, which launched the job before boot completed. New logic
waits for the boot to complete and slurmd to register on the node
before clearing the PowerUp flag.

Previous logic was not waiting for nodes to all go into "on" state

Found with -Werror=old-style-declaration, remove it rather than move
'inline' to the start of the declaration to silence the warning.

Default value is 1. Weight is uint32_t so this check was always succeeding.

Moe noted in 2008 that this was not needed on recent kernels.
Eight years seems like enough time.

Coverity issue 70420.

Remove the unused PMI2U_FUNC macro rather than changing it.

No functional changes.

Refactor resource selection logic to support the node_feature
  AllowUserBoot configuration parameter

No functional change.

Add note to slurm.conf man page about setting "--cpu_bind=no" as part
of SallocDefaultCommand if a TaskPlugin is in use.

Test 14.10 in the test suite (of slurm 15.08.8, at least) uses

  $sinfo -tidle -h -o%n

to find idle nodes.  This only works if NodeHostname == NodeName on the
nodes.  The following should work regardless of this:

  $scontrol show hostnames \$($sinfo -tidle -h -o%N)

Revert call to getaddrinfo, restoring gethostbyaddr (introduced in Slurm
    16.05.0pre1) which was failing on some systems. Specifically test7.2
    was failing on some systems with getaddrinfo() returning an error of
    "System error: Resource temporarily unavailable". Partial reversion
    of commit 89621f65

Since the function is inlined the single definition let GCC build everything
properly, but debug builds (which disable inline) resulted in:
slurmstepd: [465.0]: symbol lookup error:
(trimmed path)/task_cgroup.so: undefined symbol: val_to_char
when running srun --cpu_bind=v.

task/affinity had this definition already, task/cgroup didn't.