Postgres-R: primary key patches

Hi,

as you might know, Postgres-R relies on primary keys to address tuples
of a table. It cannot replicate tables without a primary key.

Primary keys currently aren't really used within the executor, so I had
to extended and modify Postgres here and there, to get the required
information. To ease reviewing I have split out these modifications and
present them here as two separate little patches.

The first one, get_pkey_index_oid.diff, changes the function
relationHasPrimaryKey into GetPrimaryKeyIndexOid, which now returns an
index oid instead of just a boolean. It works pretty much the same,
except from returning an oid instead of just a boolean. (In the current
Postgres-R code, I've duplicated that code to
src/backend/replication/recovery.c)

And secondly, the add_pkey_info.diff patch adds a boolean field
ii_Primary to the IndexInfo struct and ri_PrimaryKey to the
ResultRelInfo struct, which is an index into the indexInfoArray.

I think these are relatively trivial modifications which could be
helpful for other purposes as well. So I suggest to apply them to
mainline whenever appropriate (read: choose the appropriate commit fest).

This also raises the more general question of how to start collaborating
on Postgres-R. I realize that it's a pretty huge project. However, I'm
unsure on how to ease reviewing for others, so if you have any ideas or
questions, please don't hesitate to ask.

Regards

Markus