This handles the case where some gres of a given Name have CPUs
configured for binding and others do not or have bad values.
It also fixes a bug in the conversion of that information to
a bitmap if the number of bits set is equal to the core count.

this make it easier to keep gres/gpu and gres/mps in sync

If gres.conf for mps has a Count, but no File specification, then
  evenly distributed that Count over all configured gres/gpu.
Also if gres/mps have a File specification and there is no gres/gpu
  with that same File specification, then report an error and
  discard that gres/mps record.

copy/paste error

It fails consistently on some distro

Add size of bitmaps to state logging

Add some functions to handle common logic and make it more generic
Fix some comments

if the gres.conf file record for a gres/gpu lacks a File specification
then log an error. Previously the record was silently dropped.

Bug 6229

Bug 6239

bug 6197

The previous select/cons_tres logic would under some circumstances
allocate a job more cores/CPUs than requested. One specific example
is a cluster having nodes with 4 cores, 2 hyperthreads each. A
job request for 20 tasks would launch 8 tasks on 8 CPUs on each
of 2 nodes and 4 tasks on 6 CPUs on a third node (i.e. a total of
22 CPUs when only 20 are needed).

# Conflicts:
#	src/common/gres.c

backport of 03ada72e

Bug 5682

Bug 5682 and 4584

Backport of ba07a6e09b6071

The commit introduced other problems and still needs more work.

other than knl.

Bug 5603

The previous select/cons_tres logic would under some circumstances
allocate a job more cores/CPUs than requested. One specific example
is a cluster having nodes with 4 cores, 2 hyperthreads each. A
job request for 20 tasks would launch 8 tasks on 8 CPUs on each
of 2 nodes and 4 tasks on 6 CPUs on a third node (i.e. a total of
22 CPUs when only 20 are needed).

Only use removed prior to 17.11 in 04bc96f6.

Bug 6261.

Bug 5603

This patch also simplifies the code using bit_cnt instead of
_bitstr_bits().  This isn't really necessary, but does look cleaner.

Bug 6216

Bug 6420

without this logic the job could get more gres on a node than there are
CPUs for

Continuation of 2 commits ago.

Bug 6159

Continuation of last commit

Bug 6159

It was using the local cluster, sending the query to the wrong table

Bug 6159

Update GRES docs regarding ignore records and links.
Update tests to get rid of extra fields for GRES ignores lines.
Add tests to check for improved ignore syntax.
Test the gres.conf examples.
Add mps to regression slurm.conf, to test mps record parsing.

Bug 5520

Add --version option to slurmd, and document that new option.

This has been disabled since before Slurm 1.0.0.

No functional change, all these are in comments.

Insure that output appears in a fixed order for parsing by test.