uglification/wrapping for parallel mode sources

[Benjamin Kosnik <bkoz at redhat dot com>]

I've gotten some feedback about the formatting of the parallel sources.
In particular, the lack of 80 column wrapping and the lack of
uglification. Both should be fixed.

I'd actually played with compressing some of the current names for
template arguments, based on some of the recent checkins to
stl_algobase.h, in particular the __lc bits. 

    template<typename _RAI1, typename _RAI2>
        static _RAI1
        __newlast1(_RAI1 __first1, _RAI1 __last1,
		   _RAI2 __first2, _RAI2 __last2)

ends up being easier to uglify and wrap than:

    template<typename _RandomAcessIterator1, typename
_RandomAccessIterator2> static _RandomAccessIterator1
        __newlast1(_RandomAccessIterator1 __first1,
_RandomAccessIterator1 __last1, _RandomAccessIterator2 __first2,
_RandomAccessIterator2 __last2)

So, I'd like to use this shortened method consistently, throughout
libstdc++.

You can see some of the shorted template argument names in

bits/algorithfwd.h
parallel/algorithmfwd.h

etc. 

Thoughts?

-benjamin