Say in the error msg that not only might the file be writable by other,
but also by group.  Kinda deceiving.

Merge branch 'master' of git-public.flux.utah.edu:/flux/git/users/wbsun/emulab-devel into weibin-merge

When running from suexec, no need to give full path to the minary,
or to re-direct stdout and stderr.

Limit recursive resolution to internal hosts -- so, don't let external
view clients resolve recursively.

This is the big osload_new commit.  It's not enabled by default; you have
to enable the NewOsload feature to get access to it.  Moreover, it
requires the NewOsSetup feature.

Conflicts:
	db/EmulabConstants.pm.in

File::stat doesn't work; CORE::stat does, so switch out File::stat.

Bugfixes for access_keys stuff -- syntax!

Add pcRemote stub class; just inherits from pc class.

Mostly some missing dependencies that turned up during our upgrade.

Now that we're entering the type-specific osload module era, it's
conceivable that modules might expose different parameters that users
could set (i.e., choose to reload via xmodem or tftp).  So, we allow users
to invoke os_load with a list of key/value pairs that get placed into the
reload_args hash argument to osload().

This will probably change later; it just preserves API compat for now.

Sometimes we want to mark some node types as not rebootable by default,
but we end up needing to power cycle them occasionally.  For instance, if
we're going to be flashing switches, we don't want any other Emulab
service power cycling them if they are in some unknown state -- only the
service flashing them should be able to power cycle them, because only it
knows what state the node is in.

Tracking node operational state across daemons (or just across fork
boundaries) is a more complex project, and it's not clear it's really
needed, aside from special cases like flashing switches.

So, we just do a little end-around the rebootable attribute.

Allow xmodem reloads if explicitly specified requested via os_load.

Make power cycles asynch; we want to track their outcome via the console.

Make sure to reload switches that have no entries in the partitions
table too.

Misc bugfixes and reorganization.

Add back in widearea support by setting the access_key field for each
image that is going to be loaded on a widearea node.