[tabled patch 0/5] Metadata replication test

Jeff, I'm thinking it may be good if you only accept 4 of these.

The test (patch #5) has two major problems - aside from the fact that
it is incomplete:

 - The libdb-5.0.26 problem: I found that perfectly good code that
   as far as I can tell conforms to Oracle docs fails on db5.
   Works fine on db4.8. No idea what's wrong. Oracle gives me
   a runaround with it

   http://forums.oracle.com/forums/thread.jspa?threadID=1110189&tstart=0

   Therefore, if we enable the patch #5, we're setting tabled to
   fail when built in Rawhide.

 - Everything looks damn kludgy and littered with magic delays.
   I think maybe some kind of good restructuring is in order.

But please feel free to offer advice if any.

-- Pete