Add support for fast-math optimizations

[elliottslaughter]

Adds support for fast-math optimizations in terralib.saveobj.

Fixes #530

TODO:

• documentation
• cudalib.toptx
• choose default fast-math options on GPU? (going to keep optimizations disabled by default, users can enable them)

[elliottslaughter]

This is pretty much ready to go. The one remaining question is what default to set for (particularly GPU) fast-math flags.

Originally, Terra used NVVM for CUDA code generation. (We still support this path with LLVM 3.8.) It turns out that NVVM performs the equivalent of LLVM's contract fast-math flag by default. In fact, I'm not aware of any way to turn this off. Strictly speaking, this is incorrect, but the cost must be small enough that NVIDIA decided it was worth it. They made the same decision in NVCC, which I guess is not surprising.

With the migration to NVPTX, this goes away. Fast-math flags are encoded in each floating-point LLVM instruction, and NVPTX actually respects these (as opposed to NVVM which applies contract whether users asked for it or not). Because the default is to have all flags disabled, this means that users on NVPTX (and therefore, from Terra's perspective, all users on LLVM 5.0 or newer) see a performance regression via NVVM. But it is strictly speaking more correct. In other words, NVVM's choice might introduce some numerical inaccuracy, and NVPTX's default avoids that (at the cost of performance when this is not necessary).

Since it's easy to turn these flags on now, I think it's reasonable to leave them off by default, but again this does result in a performance regression for anyone who has been sticking it out with LLVM 3.8 and doesn't pay attention to the new arguments to terralib.saveobj / cudalib.toptx.

Edit: to be clear, the option that has been implemented in this PR is to keep fast-math flags disabled by default.

If anyone has opinions on this, let me know.

[elliottslaughter]

CI failure looks like a Nix issue, and my last (nearly identical) commit passed, so I'm going to go ahead and merge this.