Reasonable NFS systems do not need a minute to propagate changes.

In favor of just using the -a option to show the tracking federated
jobs. This allows scontrol -a show jobs to show the tracking jobs as
well.

to indicate wheter the job was requeue held or not. This enables the
federation to trigger off whether the job was requeue held or not.

So that the origin job tell a remote cluster to cancel the job but mark
the job as requeued in the database.

See note about the KILL_* flags actually using 12bits instead of noted
8bits.

Follows pattern from c5ace562

If a job was requeued while in the completing state, the database wasn't
being updated with the requeue state.

When a fed job is requeued, it needs to be requeued to clusters that it was
submittted to.

When the a fed job is requeued and new siblings are submitted to the
other siblings, the restart_cnt needs to go to the siblings in case the
job runs on a remote sibling.

The federation needs to make a job_desc when requeueing jobs to
siblings.

Since a persistent connection can only be established by SlurmUser this
prevents non-slurmuser users from calling the rpcs. It also requires that all
slurmctlds in the federation have the same SlurmUser.

If the job can't start now, just submit the job to all siblings.

_update_sibling_job_siblings()

like it does in slurm_send_recv_msg. The resp needs to be inited before
_check_send it called.

Sibling jobs have to get lock from the origin cluster in order to attempt to
allocate nodes. If it gets the allocation then it lets the origin cluster know
and the origin cluster will set the siblings jobs, if any, into a REVOKED state
and purge the jobs. If the sibling job is the only sibling then it assumes the
lock and attempts to start the job to avoid extra communications. If nodes can't
be allocated then the job releases the lock for another cluster to try.