Continuations of 86ed603c
CID: 161500, 45176

No functional change.

Bug 3447

Makes it so that an srun/salloc initiated from a cluster with a one
switch type will work on a different cluster with a different switch
type.

link to libslurm as well as libslurmdb to make things work.

dynamically or statically to libslurm.

This can result in much smaller binaries, but isn't the easiest to develop against as if the files in
src/common or src/*api change and the binaries linking against them are compiled directly
libslurm.so doesn't get recompiled automatically, you have to manually do it.  But in production
this could give large benefits as Slurm's footprint is now considerably smaller.

This used to not work on environments like AIX where plugins couldn't resolve variables used in
the parent program, but this appears to not be the case now we link to the "full" .so which exports
everything as the .o does.

Dead code, since xmalloc will abort() on OOM.

Bug 3444

Remove a couple of unnecesaary fatal() calls and remove some other
  unnecessary logic.

This would not really be a fatal condition (although we would
  eventually exhaust memory

This would not really be a fatal condition

bit_alloc() will abort if the malloc fails

All of these fatal conditions are already handled by xmalloc()

These were false positive error reports, but should go away now.

These are all false positives, but code will remove the Coverity
  error reports

Add upper bounds to elimate "Untrusted loop bound" errors.

Passes the client's origin cluster name to the controller. If the
client's cluster name is different from the controller's cluster name it
will send back the cluster and node's information for the salloc/srun to
talk to.
e.g.
cluster1$ srun -Mcluster2 hostname

if present.

Previously it was just putting the IP and the srun would check to see if
slurm.conf's cluster name and the IP were the same to prevent setting
the working cluster unless if needed to. But since it was was checking
the cluster name and the IP it would always try to set the working
cluster rec even if they were the same.

See if this tactic can be used to make a bunch of Coverity errors
  go away.

Add NO_VAL32 (same value as NO_VAL) and
add INFINITE32 (same value as INFINITE)
These now match the NO_VAL16, NO_VAL64, INFINITE16 and INFINITE64
value types

Coverity does not recongize the getopt_long() function's required
  argument, so Coverity reported errors whenever a NULL optarg
  value resulted in an error. This change adds a bunch of tests
  for NULL optarg value and skips it's processing.

Bug 3406.

Bug 3122.