This is needed because if a configure option has " as its value,
it will be rendered as {"}; after stripping one level of quoting it's
just ", on which splitlist will barf.

checking nits.

(testall.py): added test_thread to the list of regression tests.

	<leonard@dstc.edu.au>; allows hyphen and period in the middle
	of attribute names.  Still not allowed as first character;
	as first character these are illegal in the Reference Concrete
	Syntax, and we've not identified any use of these characters as
	the first char in an attribute name in deployment on the web.

SIGCPU and SIGFSZ but we're (Jeremy and I) are actually unsure whether
these were typos or if there are systems that use these alternate
names.  We've checked Solaris, SunOS, and IRIX; they contain only the
SIGX* names.

	(For those watching Python CVS messages:  I checked in all of
	Tamito's other changes, but the mail delivery failed since the
	subject line was too long.  The patch Tamito sent for the
	documentation has been completely merged in.)

	Merged in many typo corrections and fixes to support GNU info
	processing; submitted by Tamito Kajiyama.

	per usage needed for GNU info processing.  Based on comments from
	Tamito Kajiyama.

	GNU info processing.

I've attempted to make a test that silently exits if either
module dl is not present, we're not on a Sun OS, or a standard
shared library ('/usr/lib/libresolv.so') is not found...  Otherwise,
It does a simple test of dlmodule on that library.  I *think*
this would be ok to add to testall.py but I'll wait till I hear
some feedback on the liberalness of this approach.

don't need the if test for verbosity when checking for results of an
unsupported option.

running verbose.

1. If a conversion isn't supported on the current platform, just
   ignore it, unless running as a script (i.e. verbose)

2. Don't use time.time() and os.popen('date') to get the raw values.
   These will always be different!