A test for replay of regression tests

Hi,

We have automated tests for many specific replication and recovery
scenarios, but nothing that tests replay of a wide range of records.
People working on recovery code often use installcheck (presumably
along with other custom tests) to exercise it, sometimes with
wal_consistency_check enabled.  So, why don't we automate that?  Aside
from exercising the WAL decoding machinery (which brought me here),
that'd hopefully provide some decent improvements in coverage of the
various redo routines, many of which are not currently exercised at
all.

I'm not quite sure where it belongs, though.  The attached initial
sketch patch puts it under rc/test/recovery near other similar things,
but I'm not sure if it's really OK to invoke make -C ../regress from
here.  I copied pg_update/test.sh's trick of using a different
outputdir to avoid clobbering a concurrent run under src/test/regress,
and I also needed to invent a way to stop it from running the cursed
tablespace test (deferring startup of the standby also works but eats
way too much space, which I learned after blowing out a smallish
development VM's disk).   Alternatively, we could put it under
src/test/regress, which makes some kind of logical sense, but it would
make a quick "make check" take more than twice as long.  Perhaps it
could use a different target, one that check-world somehow reaches but
not check, and that also doesn't seem great.  Ideas on how to
structure this or improve the perl welcome.