Another buggy complex workaround

[Gabriel Dos Reis]

petter.urkedal@telia.com writes:

| The library can not be used for complex math at the time, since
| gcc gives an ICE under some circumstances when it has to synthesize
| the copy-ctors of the complex<float>, complex<double> and
| complex<long double> classes.  (The library compiles fine, though).
| 
| The fix is simply to define these constructors.  Attached is a patch
| similar to the workaround for __complex__ float done by Mumit Khan
| 1999-11-21.  Does this fix the __complex__ float bug, as well?
| 
| 2000-02-05  Petter Urkedal  <petter@matfys.lth.se>
| 
|         * acinclude.m4: New GLIBCPP_CHECK_COMPLEX_SUPPORT macro. Based
| 	on the 1999-11-21 entries by Mumit Khan.
|         * configure.in: Use.
|         * acconfig.h: New _GLIBCPP_BUGGY_COMPLEX macro.
|         * bits/std_complex.h: Use.

Applied.

-- Gaby