Set the Module's target appropriately when it's first created.
Compile separate 32 and 64 bit versions of the builtins-c bitcocde
and load the appropriate one based on the target we're compiling
for.
Add optimization patterns to detect and simplify masked loads and stores
with the mask all on / all off.
Enable AVX for LLVM 3.0 builds (still generally hits bugs / unimplemented
stuff on the LLVM side, but it's getting there).
