Since the objects are the same, could generate SINGLE/COMPLEX/etc here
without having to rewrite all the object enumeration code again.

This function takes a set of sources and a set of preprocessor
definitions. It will iterate over the sources and build an object
file for each combination of preprocessor definitions for each
source file.

Added a rather convoluted CMake function to find all combinations
of a given list. This will be useful for the object files that are
compiled multiple times with different combinations of preprocessor
definitions.

Fix a segfault in gemv when MAX_STACK_ALLOC is set

* stack_alloc_size is needed after the implementation call
but it may be overwritten if it's optimized to a register,
because some gemv implementation (ex: dgemv_n.S) do not
restore all register (ex: r10).
* do the same in ger.c for the same reasons even if the bug
has not been observed.

Still isn't parsed into CMake variables, and getarch_2 needs to
get the same treatment.

It only does the double precision compile currently.

I realized I didn't finish converting Makefile.system yet, so I made
a note of that.

The main part of this target is looping through the BLAS subfolders
and calling make on them. Need to add CMakeLists.txt for each of these
subfolders.

Move #include statements outside extern "C" blocks

Add test for drotmg bug fixed by 692b14c

Test requested in issue xianyi/OpenBLAS#484.

Run tests by applying the following change and then make:

	diff --git a/Makefile.rule b/Makefile.rule
	index bea1fe1..9852ff3 100644
	--- a/Makefile.rule
	+++ b/Makefile.rule
	@@ -140,7 +140,7 @@ NO_AFFINITY = 1

	-# UTEST_CHECK = 1
	+UTEST_CHECK = 1

Doesn't actually run it yet.

Allow to do gemv and ger buffer allocation on the stack

Optimizations for steamroller

added Steamroller as a  cpu target

ger and gemv call blas_memory_alloc/free which in their turn
call blas_lock. blas_lock create thread contention when matrices
are small and the number of thread is high enough. We avoid
call blas_memory_alloc by replacing it with stack allocation.
This can be enabled with:
make -DMAX_STACK_ALLOC=2048
The given size (in byte) must be high enough to avoid thread contention
and small enough to avoid stack overflow.

Fix #478

Correct ilaver C declaration

workaround for sandybridge zgemm kernel