This will make error handling simpler as we want API users to not
see a fatal error/exit with a bad configuration, but only an error.
bug 374 (movement toward a fix)

On either startup or reconfiguration, log the event and set the node
state to FUTURE.

to drain the node and log error slurmd log file.

MaxBlockInError set the job wait reason appropriately.

at debug() level, make the argument check case insensitive,
avoid printing duplicate \n.

already using select/cray