With the previous behavior, any fact which depended on another fact in a file
not matching its name would fail to properly load the required fact, resulting
in missing, incorrect, or inconsistent values.

For instance, the operatingsystem fact depends on the lsbdistid fact found in
lsb.rb. The first time the operatingsystem fact is requested, it requires
lsb.rb, and so the required fact is loaded first. But if Facter is subsequently
cleared and the operatingsystem fact requested again, the require will not
occur, and the fact will not be automatically loaded because it doesn't match
its filename.

Now if a fact is requested and can't be found, we will attempt to load all the
facts to find it. Such an approach appears heavy-handed, but in the case where
we want a particular fact, this is the only way to make sure we've found it. In
the case where we eventually want other facts, we are conveniently eagerly
loading them.

Paired-With: Jacob Helwig <jacob@puppetlabs.com>

Dir.glob returns an array, but a test was stubbing it to return a
string.  In Ruby 1.8.7 if you call enumerable methods (each, collect,
etc) on a string, the string is split on \n first.  This meant the poor
stubbing didn't affect 1.8.7, but 1.9.2 is more strict and won't
automatically convert when you call enumerable methods on strings.

Paired-with: Josh Cooper <josh@puppetlabs.com>

* ticket/next/6728-facter_improperly_detects_openvzve_on_cloudlinux_systems:
  (#6728) Facter improperly detects openvzve on CloudLinux systems
  (#6883) Update Facter install.rb to be slightly more informative.

The sysctl uptime fixture was little endian, which fails on sparc
hardware. Added a big endian fixture and endian detection.

Make the openvz check for more than just the vz directory to be sure it's
OpenVZ.

Update the OpenVZ fact to be slightly more resilient in it's checking of
OpenVZ, so it doesn't clash with CloudLinux's LVE container system.

renames Facter::Util::Resolution#length to weight as a more generic
mechanism for allowing resolutions to state their importance

* adrien/tickets/next/2714:
  (#2714) Fixed faulty test
  (#2714) Added timeout to prtdiag resulution

 - Looks like some copy/paste added some linux tests that relied on
   prtdiag, which isn't possible. Corrected them to check against
   solaris.

 - Refactored code to make testing simpler