cleanup patches for incremental backup

Hi,

I discovered that my patch to add WAL summarization added two new
SQL-callable functions but failed to document them. 0001 fixes that.

An outstanding item from the original thread was to write a better
test for the not-yet-committed pg_walsummary utility. But I discovered
that I couldn't do that because there were some race conditions that
couldn't easily be cured. So 0002 therefore adds a new function
pg_get_wal_summarizer_state() which returns various items of in-memory
state related to WAL summarization. We had some brief discussion of
this being desirable for other reasons; it's nice for users to be able
to look at this information in case of trouble (say, if the summarizer
is not keeping up).

0003 then adds the previously-proposed pg_walsummary utility, with
tests that depend on 0002.

0004 attempts to fix some problems detected by Coverity and subsequent
code inspection.

--
Robert Haas
EDB: http://www.enterprisedb.com