[Aboriginal] Temporarily reverting uClibc to get a release out.

[Rob Landley]

Ok, commit 1329 back in February updated uClibc to -rc2, and then 1330
changed the baseconfig, 1338 swapped around the patches, 1339 bumped the
version to -rc3, and 1340 switched everything from setjump/longjmp stack
unwinding to dwarf (because NPTL requires it).

NPTL doesn't even build for x86-64 for me because it's trying to suck
non-PIC binary code into a shared library.  I'm not sure why this works
for other people, haven't tracked it down yet.  But nobody on the list
was interested in the issue:

  http://lists.uclibc.org/pipermail/uclibc/2011-March/045020.html

Switching back to pthreads makes most of hte targets work, but powerpc
is still broken, dunno why.

Attempting to build LFS under the result (on target i686) runs into fun
little issues like:

  http://lists.uclibc.org/pipermail/uclibc/2011-March/045021.html

To which the uClibc developers basically said the fact that this package
used to build unmodified against uClibc, and no longer does, is not
something they consider a regression.

Uh-huh.

So for right now, I need to untangle those 5 commits above, put the old
uClibc back, and move the -rc series into the USE_UNSTABLE slot so I can
keep testing/fixing it no matter what its upstream developers think.
(And the fact that the packages are alt-uClibc but the trigger is
USE_UNSTABLE=uClibc is an annoyance I need to resolve.  Consistent
naming is a good thing, hard to even talk about it otherwise...)

Rob