pgindent

I think it's about time for us to run pgindent. I did a trial run
today of pgindent today and came up with the attached patch for
typedefs.list, which I'd like to commit more or less immediately,
barring objections. It mostly just adds new typedefs that have
appeared over the last year, but it also realphabetizes the file -
some things that were added incrementally seem to have ended up in
what is, at least according to what sort likes to do on my machine,
the wrong place in the file.

With this applied, I get a fairly clean pgindent run. There are some
problems with comments getting mangled, and in a couple of cases
function definitions getting mangled, that need more investigation.
I'll try to find time to look into that soon and follow up, unless
somebody else beats me to it. As far as possible, I think it's
desirable to clean up those things before rather than after running
pgindent, because unmangling ASCII art that pgindent has stepped on is
a thankless chore.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company